Search Results

Search found 3996 results on 160 pages for 'operations'.

Page 16/160 | < Previous Page | 12 13 14 15 16 17 18 19 20 21 22 23  | Next Page >

  • Is this a good approach to execute a list of operations on a data structure in Python?

    - by Sridhar Iyer
    I have a dictionary of data, the key is the file name and the value is another dictionary of its attribute values. Now I'd like to pass this data structure to various functions, each of which runs some test on the attribute and returns True/False. One approach would be to call each function one by one explicitly from the main code. However I can do something like this: #MYmodule.py class Mymodule: def MYfunc1(self): ... def MYfunc2(self): ... #main.py import Mymodule ... #fill the data structure ... #Now call all the functions in Mymodule one by one for funcs in dir(Mymodule): if funcs[:2]=='MY': result=Mymodule.__dict__.get(funcs)(dataStructure) The advantage of this approach is that implementation of main class needn't change when I add more logic/tests to MYmodule. Is this a good way to solve the problem at hand? Are there better alternatives to this solution?

    Read the article

  • When machine code is generated from a program how does it translates to hardware level operations ??

    - by user553492
    Like if say the instruction is something like 100010101 1010101 01010101 011101010101. Now how is this translating to an actual job of deleting something from memory? Memory consists of actual physical transistors the HOLD data. What causes them to lose that data is some external signal? I want to know how that signal is generated. Like how some binary numbers change the state of a physical transistor. Is there a level beyond machine code that isn't explicitly visible to a programmer? I have heard of microcode that handle code at hardware level, even below assembly language. But still I pretty much don't understand. Thanks!

    Read the article

  • What PHP function(s) can I use to perform operations on non-integer timestamps?

    - by stephenhay
    Disclaimer, I'm not a PHP programmer, so you might find this question trivial. That's why I'm asking you! I've got this kind of timestamp: 2010-05-10T22:00:00 (That's Y-m-d) I would like to subtract, say, 10 days (or months, whatever) from this, and have my result be in the same format, i.e. 2010-04-30T22:00:00. What function(s) do I need to do this in PHP? Note: I'm using this to do a computed field in Drupal. The result will be the date that an e-mail is sent. Bonus question: If 2010-05-10T22:00:00 means "May 10, 2010 at 10pm", is there a timestamp equivalent of "May 10, 2010 (all day)"? Thanks everyone.

    Read the article

  • Are there any javascript string formatting operations similar to the way %s is used in Python?

    - by Phil
    I've been writing a lot of javascript, and when I want to stick a variable in a string, I've been doing it like so: $("#more_info span#author").html("Created by: <a href='/user/" + author + "'>" + author + "</a>"); I feel like it's pretty ugly and a pain to write over and over. In python the %s operator makes this problem easy. Even in C, I can do sprintf (IIRC). Is there anything like that in javascript? (Lots of google'ing yielded nothing.)

    Read the article

  • How John Got 15x Improvement Without Really Trying

    - by rchrd
    The following article was published on a Sun Microsystems website a number of years ago by John Feo. It is still useful and worth preserving. So I'm republishing it here.  How I Got 15x Improvement Without Really Trying John Feo, Sun Microsystems Taking ten "personal" program codes used in scientific and engineering research, the author was able to get from 2 to 15 times performance improvement easily by applying some simple general optimization techniques. Introduction Scientific research based on computer simulation depends on the simulation for advancement. The research can advance only as fast as the computational codes can execute. The codes' efficiency determines both the rate and quality of results. In the same amount of time, a faster program can generate more results and can carry out a more detailed simulation of physical phenomena than a slower program. Highly optimized programs help science advance quickly and insure that monies supporting scientific research are used as effectively as possible. Scientific computer codes divide into three broad categories: ISV, community, and personal. ISV codes are large, mature production codes developed and sold commercially. The codes improve slowly over time both in methods and capabilities, and they are well tuned for most vendor platforms. Since the codes are mature and complex, there are few opportunities to improve their performance solely through code optimization. Improvements of 10% to 15% are typical. Examples of ISV codes are DYNA3D, Gaussian, and Nastran. Community codes are non-commercial production codes used by a particular research field. Generally, they are developed and distributed by a single academic or research institution with assistance from the community. Most users just run the codes, but some develop new methods and extensions that feed back into the general release. The codes are available on most vendor platforms. Since these codes are younger than ISV codes, there are more opportunities to optimize the source code. Improvements of 50% are not unusual. Examples of community codes are AMBER, CHARM, BLAST, and FASTA. Personal codes are those written by single users or small research groups for their own use. These codes are not distributed, but may be passed from professor-to-student or student-to-student over several years. They form the primordial ocean of applications from which community and ISV codes emerge. Government research grants pay for the development of most personal codes. This paper reports on the nature and performance of this class of codes. Over the last year, I have looked at over two dozen personal codes from more than a dozen research institutions. The codes cover a variety of scientific fields, including astronomy, atmospheric sciences, bioinformatics, biology, chemistry, geology, and physics. The sources range from a few hundred lines to more than ten thousand lines, and are written in Fortran, Fortran 90, C, and C++. For the most part, the codes are modular, documented, and written in a clear, straightforward manner. They do not use complex language features, advanced data structures, programming tricks, or libraries. I had little trouble understanding what the codes did or how data structures were used. Most came with a makefile. Surprisingly, only one of the applications is parallel. All developers have access to parallel machines, so availability is not an issue. Several tried to parallelize their applications, but stopped after encountering difficulties. Lack of education and a perception that parallelism is difficult prevented most from trying. I parallelized several of the codes using OpenMP, and did not judge any of the codes as difficult to parallelize. Even more surprising than the lack of parallelism is the inefficiency of the codes. I was able to get large improvements in performance in a matter of a few days applying simple optimization techniques. Table 1 lists ten representative codes [names and affiliation are omitted to preserve anonymity]. Improvements on one processor range from 2x to 15.5x with a simple average of 4.75x. I did not use sophisticated performance tools or drill deep into the program's execution character as one would do when tuning ISV or community codes. Using only a profiler and source line timers, I identified inefficient sections of code and improved their performance by inspection. The changes were at a high level. I am sure there is another factor of 2 or 3 in each code, and more if the codes are parallelized. The study’s results show that personal scientific codes are running many times slower than they should and that the problem is pervasive. Computational scientists are not sloppy programmers; however, few are trained in the art of computer programming or code optimization. I found that most have a working knowledge of some programming language and standard software engineering practices; but they do not know, or think about, how to make their programs run faster. They simply do not know the standard techniques used to make codes run faster. In fact, they do not even perceive that such techniques exist. The case studies described in this paper show that applying simple, well known techniques can significantly increase the performance of personal codes. It is important that the scientific community and the Government agencies that support scientific research find ways to better educate academic scientific programmers. The inefficiency of their codes is so bad that it is retarding both the quality and progress of scientific research. # cacheperformance redundantoperations loopstructures performanceimprovement 1 x x 15.5 2 x 2.8 3 x x 2.5 4 x 2.1 5 x x 2.0 6 x 5.0 7 x 5.8 8 x 6.3 9 2.2 10 x x 3.3 Table 1 — Area of improvement and performance gains of 10 codes The remainder of the paper is organized as follows: sections 2, 3, and 4 discuss the three most common sources of inefficiencies in the codes studied. These are cache performance, redundant operations, and loop structures. Each section includes several examples. The last section summaries the work and suggests a possible solution to the issues raised. Optimizing cache performance Commodity microprocessor systems use caches to increase memory bandwidth and reduce memory latencies. Typical latencies from processor to L1, L2, local, and remote memory are 3, 10, 50, and 200 cycles, respectively. Moreover, bandwidth falls off dramatically as memory distances increase. Programs that do not use cache effectively run many times slower than programs that do. When optimizing for cache, the biggest performance gains are achieved by accessing data in cache order and reusing data to amortize the overhead of cache misses. Secondary considerations are prefetching, associativity, and replacement; however, the understanding and analysis required to optimize for the latter are probably beyond the capabilities of the non-expert. Much can be gained simply by accessing data in the correct order and maximizing data reuse. 6 out of the 10 codes studied here benefited from such high level optimizations. Array Accesses The most important cache optimization is the most basic: accessing Fortran array elements in column order and C array elements in row order. Four of the ten codes—1, 2, 4, and 10—got it wrong. Compilers will restructure nested loops to optimize cache performance, but may not do so if the loop structure is too complex, or the loop body includes conditionals, complex addressing, or function calls. In code 1, the compiler failed to invert a key loop because of complex addressing do I = 0, 1010, delta_x IM = I - delta_x IP = I + delta_x do J = 5, 995, delta_x JM = J - delta_x JP = J + delta_x T1 = CA1(IP, J) + CA1(I, JP) T2 = CA1(IM, J) + CA1(I, JM) S1 = T1 + T2 - 4 * CA1(I, J) CA(I, J) = CA1(I, J) + D * S1 end do end do In code 2, the culprit is conditionals do I = 1, N do J = 1, N If (IFLAG(I,J) .EQ. 0) then T1 = Value(I, J-1) T2 = Value(I-1, J) T3 = Value(I, J) T4 = Value(I+1, J) T5 = Value(I, J+1) Value(I,J) = 0.25 * (T1 + T2 + T5 + T4) Delta = ABS(T3 - Value(I,J)) If (Delta .GT. MaxDelta) MaxDelta = Delta endif enddo enddo I fixed both programs by inverting the loops by hand. Code 10 has three-dimensional arrays and triply nested loops. The structure of the most computationally intensive loops is too complex to invert automatically or by hand. The only practical solution is to transpose the arrays so that the dimension accessed by the innermost loop is in cache order. The arrays can be transposed at construction or prior to entering a computationally intensive section of code. The former requires all array references to be modified, while the latter is cost effective only if the cost of the transpose is amortized over many accesses. I used the second approach to optimize code 10. Code 5 has four-dimensional arrays and loops are nested four deep. For all of the reasons cited above the compiler is not able to restructure three key loops. Assume C arrays and let the four dimensions of the arrays be i, j, k, and l. In the original code, the index structure of the three loops is L1: for i L2: for i L3: for i for l for l for j for k for j for k for j for k for l So only L3 accesses array elements in cache order. L1 is a very complex loop—much too complex to invert. I brought the loop into cache alignment by transposing the second and fourth dimensions of the arrays. Since the code uses a macro to compute all array indexes, I effected the transpose at construction and changed the macro appropriately. The dimensions of the new arrays are now: i, l, k, and j. L3 is a simple loop and easily inverted. L2 has a loop-carried scalar dependence in k. By promoting the scalar name that carries the dependence to an array, I was able to invert the third and fourth subloops aligning the loop with cache. Code 5 is by far the most difficult of the four codes to optimize for array accesses; but the knowledge required to fix the problems is no more than that required for the other codes. I would judge this code at the limits of, but not beyond, the capabilities of appropriately trained computational scientists. Array Strides When a cache miss occurs, a line (64 bytes) rather than just one word is loaded into the cache. If data is accessed stride 1, than the cost of the miss is amortized over 8 words. Any stride other than one reduces the cost savings. Two of the ten codes studied suffered from non-unit strides. The codes represent two important classes of "strided" codes. Code 1 employs a multi-grid algorithm to reduce time to convergence. The grids are every tenth, fifth, second, and unit element. Since time to convergence is inversely proportional to the distance between elements, coarse grids converge quickly providing good starting values for finer grids. The better starting values further reduce the time to convergence. The downside is that grids of every nth element, n > 1, introduce non-unit strides into the computation. In the original code, much of the savings of the multi-grid algorithm were lost due to this problem. I eliminated the problem by compressing (copying) coarse grids into continuous memory, and rewriting the computation as a function of the compressed grid. On convergence, I copied the final values of the compressed grid back to the original grid. The savings gained from unit stride access of the compressed grid more than paid for the cost of copying. Using compressed grids, the loop from code 1 included in the previous section becomes do j = 1, GZ do i = 1, GZ T1 = CA(i+0, j-1) + CA(i-1, j+0) T4 = CA1(i+1, j+0) + CA1(i+0, j+1) S1 = T1 + T4 - 4 * CA1(i+0, j+0) CA(i+0, j+0) = CA1(i+0, j+0) + DD * S1 enddo enddo where CA and CA1 are compressed arrays of size GZ. Code 7 traverses a list of objects selecting objects for later processing. The labels of the selected objects are stored in an array. The selection step has unit stride, but the processing steps have irregular stride. A fix is to save the parameters of the selected objects in temporary arrays as they are selected, and pass the temporary arrays to the processing functions. The fix is practical if the same parameters are used in selection as in processing, or if processing comprises a series of distinct steps which use overlapping subsets of the parameters. Both conditions are true for code 7, so I achieved significant improvement by copying parameters to temporary arrays during selection. Data reuse In the previous sections, we optimized for spatial locality. It is also important to optimize for temporal locality. Once read, a datum should be used as much as possible before it is forced from cache. Loop fusion and loop unrolling are two techniques that increase temporal locality. Unfortunately, both techniques increase register pressure—as loop bodies become larger, the number of registers required to hold temporary values grows. Once register spilling occurs, any gains evaporate quickly. For multiprocessors with small register sets or small caches, the sweet spot can be very small. In the ten codes presented here, I found no opportunities for loop fusion and only two opportunities for loop unrolling (codes 1 and 3). In code 1, unrolling the outer and inner loop one iteration increases the number of result values computed by the loop body from 1 to 4, do J = 1, GZ-2, 2 do I = 1, GZ-2, 2 T1 = CA1(i+0, j-1) + CA1(i-1, j+0) T2 = CA1(i+1, j-1) + CA1(i+0, j+0) T3 = CA1(i+0, j+0) + CA1(i-1, j+1) T4 = CA1(i+1, j+0) + CA1(i+0, j+1) T5 = CA1(i+2, j+0) + CA1(i+1, j+1) T6 = CA1(i+1, j+1) + CA1(i+0, j+2) T7 = CA1(i+2, j+1) + CA1(i+1, j+2) S1 = T1 + T4 - 4 * CA1(i+0, j+0) S2 = T2 + T5 - 4 * CA1(i+1, j+0) S3 = T3 + T6 - 4 * CA1(i+0, j+1) S4 = T4 + T7 - 4 * CA1(i+1, j+1) CA(i+0, j+0) = CA1(i+0, j+0) + DD * S1 CA(i+1, j+0) = CA1(i+1, j+0) + DD * S2 CA(i+0, j+1) = CA1(i+0, j+1) + DD * S3 CA(i+1, j+1) = CA1(i+1, j+1) + DD * S4 enddo enddo The loop body executes 12 reads, whereas as the rolled loop shown in the previous section executes 20 reads to compute the same four values. In code 3, two loops are unrolled 8 times and one loop is unrolled 4 times. Here is the before for (k = 0; k < NK[u]; k++) { sum = 0.0; for (y = 0; y < NY; y++) { sum += W[y][u][k] * delta[y]; } backprop[i++]=sum; } and after code for (k = 0; k < KK - 8; k+=8) { sum0 = 0.0; sum1 = 0.0; sum2 = 0.0; sum3 = 0.0; sum4 = 0.0; sum5 = 0.0; sum6 = 0.0; sum7 = 0.0; for (y = 0; y < NY; y++) { sum0 += W[y][0][k+0] * delta[y]; sum1 += W[y][0][k+1] * delta[y]; sum2 += W[y][0][k+2] * delta[y]; sum3 += W[y][0][k+3] * delta[y]; sum4 += W[y][0][k+4] * delta[y]; sum5 += W[y][0][k+5] * delta[y]; sum6 += W[y][0][k+6] * delta[y]; sum7 += W[y][0][k+7] * delta[y]; } backprop[k+0] = sum0; backprop[k+1] = sum1; backprop[k+2] = sum2; backprop[k+3] = sum3; backprop[k+4] = sum4; backprop[k+5] = sum5; backprop[k+6] = sum6; backprop[k+7] = sum7; } for one of the loops unrolled 8 times. Optimizing for temporal locality is the most difficult optimization considered in this paper. The concepts are not difficult, but the sweet spot is small. Identifying where the program can benefit from loop unrolling or loop fusion is not trivial. Moreover, it takes some effort to get it right. Still, educating scientific programmers about temporal locality and teaching them how to optimize for it will pay dividends. Reducing instruction count Execution time is a function of instruction count. Reduce the count and you usually reduce the time. The best solution is to use a more efficient algorithm; that is, an algorithm whose order of complexity is smaller, that converges quicker, or is more accurate. Optimizing source code without changing the algorithm yields smaller, but still significant, gains. This paper considers only the latter because the intent is to study how much better codes can run if written by programmers schooled in basic code optimization techniques. The ten codes studied benefited from three types of "instruction reducing" optimizations. The two most prevalent were hoisting invariant memory and data operations out of inner loops. The third was eliminating unnecessary data copying. The nature of these inefficiencies is language dependent. Memory operations The semantics of C make it difficult for the compiler to determine all the invariant memory operations in a loop. The problem is particularly acute for loops in functions since the compiler may not know the values of the function's parameters at every call site when compiling the function. Most compilers support pragmas to help resolve ambiguities; however, these pragmas are not comprehensive and there is no standard syntax. To guarantee that invariant memory operations are not executed repetitively, the user has little choice but to hoist the operations by hand. The problem is not as severe in Fortran programs because in the absence of equivalence statements, it is a violation of the language's semantics for two names to share memory. Codes 3 and 5 are C programs. In both cases, the compiler did not hoist all invariant memory operations from inner loops. Consider the following loop from code 3 for (y = 0; y < NY; y++) { i = 0; for (u = 0; u < NU; u++) { for (k = 0; k < NK[u]; k++) { dW[y][u][k] += delta[y] * I1[i++]; } } } Since dW[y][u] can point to the same memory space as delta for one or more values of y and u, assignment to dW[y][u][k] may change the value of delta[y]. In reality, dW and delta do not overlap in memory, so I rewrote the loop as for (y = 0; y < NY; y++) { i = 0; Dy = delta[y]; for (u = 0; u < NU; u++) { for (k = 0; k < NK[u]; k++) { dW[y][u][k] += Dy * I1[i++]; } } } Failure to hoist invariant memory operations may be due to complex address calculations. If the compiler can not determine that the address calculation is invariant, then it can hoist neither the calculation nor the associated memory operations. As noted above, code 5 uses a macro to address four-dimensional arrays #define MAT4D(a,q,i,j,k) (double *)((a)->data + (q)*(a)->strides[0] + (i)*(a)->strides[3] + (j)*(a)->strides[2] + (k)*(a)->strides[1]) The macro is too complex for the compiler to understand and so, it does not identify any subexpressions as loop invariant. The simplest way to eliminate the address calculation from the innermost loop (over i) is to define a0 = MAT4D(a,q,0,j,k) before the loop and then replace all instances of *MAT4D(a,q,i,j,k) in the loop with a0[i] A similar problem appears in code 6, a Fortran program. The key loop in this program is do n1 = 1, nh nx1 = (n1 - 1) / nz + 1 nz1 = n1 - nz * (nx1 - 1) do n2 = 1, nh nx2 = (n2 - 1) / nz + 1 nz2 = n2 - nz * (nx2 - 1) ndx = nx2 - nx1 ndy = nz2 - nz1 gxx = grn(1,ndx,ndy) gyy = grn(2,ndx,ndy) gxy = grn(3,ndx,ndy) balance(n1,1) = balance(n1,1) + (force(n2,1) * gxx + force(n2,2) * gxy) * h1 balance(n1,2) = balance(n1,2) + (force(n2,1) * gxy + force(n2,2) * gyy)*h1 end do end do The programmer has written this loop well—there are no loop invariant operations with respect to n1 and n2. However, the loop resides within an iterative loop over time and the index calculations are independent with respect to time. Trading space for time, I precomputed the index values prior to the entering the time loop and stored the values in two arrays. I then replaced the index calculations with reads of the arrays. Data operations Ways to reduce data operations can appear in many forms. Implementing a more efficient algorithm produces the biggest gains. The closest I came to an algorithm change was in code 4. This code computes the inner product of K-vectors A(i) and B(j), 0 = i < N, 0 = j < M, for most values of i and j. Since the program computes most of the NM possible inner products, it is more efficient to compute all the inner products in one triply-nested loop rather than one at a time when needed. The savings accrue from reading A(i) once for all B(j) vectors and from loop unrolling. for (i = 0; i < N; i+=8) { for (j = 0; j < M; j++) { sum0 = 0.0; sum1 = 0.0; sum2 = 0.0; sum3 = 0.0; sum4 = 0.0; sum5 = 0.0; sum6 = 0.0; sum7 = 0.0; for (k = 0; k < K; k++) { sum0 += A[i+0][k] * B[j][k]; sum1 += A[i+1][k] * B[j][k]; sum2 += A[i+2][k] * B[j][k]; sum3 += A[i+3][k] * B[j][k]; sum4 += A[i+4][k] * B[j][k]; sum5 += A[i+5][k] * B[j][k]; sum6 += A[i+6][k] * B[j][k]; sum7 += A[i+7][k] * B[j][k]; } C[i+0][j] = sum0; C[i+1][j] = sum1; C[i+2][j] = sum2; C[i+3][j] = sum3; C[i+4][j] = sum4; C[i+5][j] = sum5; C[i+6][j] = sum6; C[i+7][j] = sum7; }} This change requires knowledge of a typical run; i.e., that most inner products are computed. The reasons for the change, however, derive from basic optimization concepts. It is the type of change easily made at development time by a knowledgeable programmer. In code 5, we have the data version of the index optimization in code 6. Here a very expensive computation is a function of the loop indices and so cannot be hoisted out of the loop; however, the computation is invariant with respect to an outer iterative loop over time. We can compute its value for each iteration of the computation loop prior to entering the time loop and save the values in an array. The increase in memory required to store the values is small in comparison to the large savings in time. The main loop in Code 8 is doubly nested. The inner loop includes a series of guarded computations; some are a function of the inner loop index but not the outer loop index while others are a function of the outer loop index but not the inner loop index for (j = 0; j < N; j++) { for (i = 0; i < M; i++) { r = i * hrmax; R = A[j]; temp = (PRM[3] == 0.0) ? 1.0 : pow(r, PRM[3]); high = temp * kcoeff * B[j] * PRM[2] * PRM[4]; low = high * PRM[6] * PRM[6] / (1.0 + pow(PRM[4] * PRM[6], 2.0)); kap = (R > PRM[6]) ? high * R * R / (1.0 + pow(PRM[4]*r, 2.0) : low * pow(R/PRM[6], PRM[5]); < rest of loop omitted > }} Note that the value of temp is invariant to j. Thus, we can hoist the computation for temp out of the loop and save its values in an array. for (i = 0; i < M; i++) { r = i * hrmax; TEMP[i] = pow(r, PRM[3]); } [N.B. – the case for PRM[3] = 0 is omitted and will be reintroduced later.] We now hoist out of the inner loop the computations invariant to i. Since the conditional guarding the value of kap is invariant to i, it behooves us to hoist the computation out of the inner loop, thereby executing the guard once rather than M times. The final version of the code is for (j = 0; j < N; j++) { R = rig[j] / 1000.; tmp1 = kcoeff * par[2] * beta[j] * par[4]; tmp2 = 1.0 + (par[4] * par[4] * par[6] * par[6]); tmp3 = 1.0 + (par[4] * par[4] * R * R); tmp4 = par[6] * par[6] / tmp2; tmp5 = R * R / tmp3; tmp6 = pow(R / par[6], par[5]); if ((par[3] == 0.0) && (R > par[6])) { for (i = 1; i <= imax1; i++) KAP[i] = tmp1 * tmp5; } else if ((par[3] == 0.0) && (R <= par[6])) { for (i = 1; i <= imax1; i++) KAP[i] = tmp1 * tmp4 * tmp6; } else if ((par[3] != 0.0) && (R > par[6])) { for (i = 1; i <= imax1; i++) KAP[i] = tmp1 * TEMP[i] * tmp5; } else if ((par[3] != 0.0) && (R <= par[6])) { for (i = 1; i <= imax1; i++) KAP[i] = tmp1 * TEMP[i] * tmp4 * tmp6; } for (i = 0; i < M; i++) { kap = KAP[i]; r = i * hrmax; < rest of loop omitted > } } Maybe not the prettiest piece of code, but certainly much more efficient than the original loop, Copy operations Several programs unnecessarily copy data from one data structure to another. This problem occurs in both Fortran and C programs, although it manifests itself differently in the two languages. Code 1 declares two arrays—one for old values and one for new values. At the end of each iteration, the array of new values is copied to the array of old values to reset the data structures for the next iteration. This problem occurs in Fortran programs not included in this study and in both Fortran 77 and Fortran 90 code. Introducing pointers to the arrays and swapping pointer values is an obvious way to eliminate the copying; but pointers is not a feature that many Fortran programmers know well or are comfortable using. An easy solution not involving pointers is to extend the dimension of the value array by 1 and use the last dimension to differentiate between arrays at different times. For example, if the data space is N x N, declare the array (N, N, 2). Then store the problem’s initial values in (_, _, 2) and define the scalar names new = 2 and old = 1. At the start of each iteration, swap old and new to reset the arrays. The old–new copy problem did not appear in any C program. In programs that had new and old values, the code swapped pointers to reset data structures. Where unnecessary coping did occur is in structure assignment and parameter passing. Structures in C are handled much like scalars. Assignment causes the data space of the right-hand name to be copied to the data space of the left-hand name. Similarly, when a structure is passed to a function, the data space of the actual parameter is copied to the data space of the formal parameter. If the structure is large and the assignment or function call is in an inner loop, then copying costs can grow quite large. While none of the ten programs considered here manifested this problem, it did occur in programs not included in the study. A simple fix is always to refer to structures via pointers. Optimizing loop structures Since scientific programs spend almost all their time in loops, efficient loops are the key to good performance. Conditionals, function calls, little instruction level parallelism, and large numbers of temporary values make it difficult for the compiler to generate tightly packed, highly efficient code. Conditionals and function calls introduce jumps that disrupt code flow. Users should eliminate or isolate conditionls to their own loops as much as possible. Often logical expressions can be substituted for if-then-else statements. For example, code 2 includes the following snippet MaxDelta = 0.0 do J = 1, N do I = 1, M < code omitted > Delta = abs(OldValue ? NewValue) if (Delta > MaxDelta) MaxDelta = Delta enddo enddo if (MaxDelta .gt. 0.001) goto 200 Since the only use of MaxDelta is to control the jump to 200 and all that matters is whether or not it is greater than 0.001, I made MaxDelta a boolean and rewrote the snippet as MaxDelta = .false. do J = 1, N do I = 1, M < code omitted > Delta = abs(OldValue ? NewValue) MaxDelta = MaxDelta .or. (Delta .gt. 0.001) enddo enddo if (MaxDelta) goto 200 thereby, eliminating the conditional expression from the inner loop. A microprocessor can execute many instructions per instruction cycle. Typically, it can execute one or more memory, floating point, integer, and jump operations. To be executed simultaneously, the operations must be independent. Thick loops tend to have more instruction level parallelism than thin loops. Moreover, they reduce memory traffice by maximizing data reuse. Loop unrolling and loop fusion are two techniques to increase the size of loop bodies. Several of the codes studied benefitted from loop unrolling, but none benefitted from loop fusion. This observation is not too surpising since it is the general tendency of programmers to write thick loops. As loops become thicker, the number of temporary values grows, increasing register pressure. If registers spill, then memory traffic increases and code flow is disrupted. A thick loop with many temporary values may execute slower than an equivalent series of thin loops. The biggest gain will be achieved if the thick loop can be split into a series of independent loops eliminating the need to write and read temporary arrays. I found such an occasion in code 10 where I split the loop do i = 1, n do j = 1, m A24(j,i)= S24(j,i) * T24(j,i) + S25(j,i) * U25(j,i) B24(j,i)= S24(j,i) * T25(j,i) + S25(j,i) * U24(j,i) A25(j,i)= S24(j,i) * C24(j,i) + S25(j,i) * V24(j,i) B25(j,i)= S24(j,i) * U25(j,i) + S25(j,i) * V25(j,i) C24(j,i)= S26(j,i) * T26(j,i) + S27(j,i) * U26(j,i) D24(j,i)= S26(j,i) * T27(j,i) + S27(j,i) * V26(j,i) C25(j,i)= S27(j,i) * S28(j,i) + S26(j,i) * U28(j,i) D25(j,i)= S27(j,i) * T28(j,i) + S26(j,i) * V28(j,i) end do end do into two disjoint loops do i = 1, n do j = 1, m A24(j,i)= S24(j,i) * T24(j,i) + S25(j,i) * U25(j,i) B24(j,i)= S24(j,i) * T25(j,i) + S25(j,i) * U24(j,i) A25(j,i)= S24(j,i) * C24(j,i) + S25(j,i) * V24(j,i) B25(j,i)= S24(j,i) * U25(j,i) + S25(j,i) * V25(j,i) end do end do do i = 1, n do j = 1, m C24(j,i)= S26(j,i) * T26(j,i) + S27(j,i) * U26(j,i) D24(j,i)= S26(j,i) * T27(j,i) + S27(j,i) * V26(j,i) C25(j,i)= S27(j,i) * S28(j,i) + S26(j,i) * U28(j,i) D25(j,i)= S27(j,i) * T28(j,i) + S26(j,i) * V28(j,i) end do end do Conclusions Over the course of the last year, I have had the opportunity to work with over two dozen academic scientific programmers at leading research universities. Their research interests span a broad range of scientific fields. Except for two programs that relied almost exclusively on library routines (matrix multiply and fast Fourier transform), I was able to improve significantly the single processor performance of all codes. Improvements range from 2x to 15.5x with a simple average of 4.75x. Changes to the source code were at a very high level. I did not use sophisticated techniques or programming tools to discover inefficiencies or effect the changes. Only one code was parallel despite the availability of parallel systems to all developers. Clearly, we have a problem—personal scientific research codes are highly inefficient and not running parallel. The developers are unaware of simple optimization techniques to make programs run faster. They lack education in the art of code optimization and parallel programming. I do not believe we can fix the problem by publishing additional books or training manuals. To date, the developers in questions have not studied the books or manual available, and are unlikely to do so in the future. Short courses are a possible solution, but I believe they are too concentrated to be much use. The general concepts can be taught in a three or four day course, but that is not enough time for students to practice what they learn and acquire the experience to apply and extend the concepts to their codes. Practice is the key to becoming proficient at optimization. I recommend that graduate students be required to take a semester length course in optimization and parallel programming. We would never give someone access to state-of-the-art scientific equipment costing hundreds of thousands of dollars without first requiring them to demonstrate that they know how to use the equipment. Yet the criterion for time on state-of-the-art supercomputers is at most an interesting project. Requestors are never asked to demonstrate that they know how to use the system, or can use the system effectively. A semester course would teach them the required skills. Government agencies that fund academic scientific research pay for most of the computer systems supporting scientific research as well as the development of most personal scientific codes. These agencies should require graduate schools to offer a course in optimization and parallel programming as a requirement for funding. About the Author John Feo received his Ph.D. in Computer Science from The University of Texas at Austin in 1986. After graduate school, Dr. Feo worked at Lawrence Livermore National Laboratory where he was the Group Leader of the Computer Research Group and principal investigator of the Sisal Language Project. In 1997, Dr. Feo joined Tera Computer Company where he was project manager for the MTA, and oversaw the programming and evaluation of the MTA at the San Diego Supercomputer Center. In 2000, Dr. Feo joined Sun Microsystems as an HPC application specialist. He works with university research groups to optimize and parallelize scientific codes. Dr. Feo has published over two dozen research articles in the areas of parallel parallel programming, parallel programming languages, and application performance.

    Read the article

  • Eclipse Error Exporting Web Project as WAR

    - by Anand
    Hi I have the following error when I export my war file org.eclipse.core.runtime.CoreException: Extended Operation failure: org.eclipse.jst.j2ee.internal.web.archive.operations.WebComponentExportOperation at org.eclipse.wst.common.frameworks.internal.datamodel.ui.DataModelWizard.performFinish(DataModelWizard.java:189) at org.eclipse.jface.wizard.WizardDialog.finishPressed(WizardDialog.java:752) at org.eclipse.jface.wizard.WizardDialog.buttonPressed(WizardDialog.java:373) at org.eclipse.jface.dialogs.Dialog$2.widgetSelected(Dialog.java:624) at org.eclipse.swt.widgets.TypedListener.handleEvent(TypedListener.java:228) at org.eclipse.swt.widgets.EventTable.sendEvent(EventTable.java:84) at org.eclipse.swt.widgets.Widget.sendEvent(Widget.java:1003) at org.eclipse.swt.widgets.Display.runDeferredEvents(Display.java:3880) at org.eclipse.swt.widgets.Display.readAndDispatch(Display.java:3473) at org.eclipse.jface.window.Window.runEventLoop(Window.java:825) at org.eclipse.jface.window.Window.open(Window.java:801) at org.eclipse.ui.internal.handlers.WizardHandler$Export.executeHandler(WizardHandler.java:97) at org.eclipse.ui.internal.handlers.WizardHandler.execute(WizardHandler.java:273) at org.eclipse.ui.internal.handlers.HandlerProxy.execute(HandlerProxy.java:294) at org.eclipse.core.commands.Command.executeWithChecks(Command.java:476) at org.eclipse.core.commands.ParameterizedCommand.executeWithChecks(ParameterizedCommand.java:508) at org.eclipse.ui.internal.handlers.HandlerService.executeCommand(HandlerService.java:169) at org.eclipse.ui.internal.handlers.SlaveHandlerService.executeCommand(SlaveHandlerService.java:241) at org.eclipse.ui.internal.actions.CommandAction.runWithEvent(CommandAction.java:157) at org.eclipse.ui.internal.actions.CommandAction.run(CommandAction.java:171) at org.eclipse.ui.actions.ExportResourcesAction.run(ExportResourcesAction.java:116) at org.eclipse.ui.actions.BaseSelectionListenerAction.runWithEvent(BaseSelectionListenerAction.java:168) at org.eclipse.jface.action.ActionContributionItem.handleWidgetSelection(ActionContributionItem.java:584) at org.eclipse.jface.action.ActionContributionItem.access$2(ActionContributionItem.java:501) at org.eclipse.jface.action.ActionContributionItem$5.handleEvent(ActionContributionItem.java:411) at org.eclipse.swt.widgets.EventTable.sendEvent(EventTable.java:84) at org.eclipse.swt.widgets.Widget.sendEvent(Widget.java:1003) at org.eclipse.swt.widgets.Display.runDeferredEvents(Display.java:3880) at org.eclipse.swt.widgets.Display.readAndDispatch(Display.java:3473) at org.eclipse.ui.internal.Workbench.runEventLoop(Workbench.java:2405) at org.eclipse.ui.internal.Workbench.runUI(Workbench.java:2369) at org.eclipse.ui.internal.Workbench.access$4(Workbench.java:2221) at org.eclipse.ui.internal.Workbench$5.run(Workbench.java:500) at org.eclipse.core.databinding.observable.Realm.runWithDefault(Realm.java:332) at org.eclipse.ui.internal.Workbench.createAndRunWorkbench(Workbench.java:493) at org.eclipse.ui.PlatformUI.createAndRunWorkbench(PlatformUI.java:149) at org.eclipse.ui.internal.ide.application.IDEApplication.start(IDEApplication.java:113) at org.eclipse.equinox.internal.app.EclipseAppHandle.run(EclipseAppHandle.java:194) at org.eclipse.core.runtime.internal.adaptor.EclipseAppLauncher.runApplication(EclipseAppLauncher.java:110) at org.eclipse.core.runtime.internal.adaptor.EclipseAppLauncher.start(EclipseAppLauncher.java:79) at org.eclipse.core.runtime.adaptor.EclipseStarter.run(EclipseStarter.java:368) at org.eclipse.core.runtime.adaptor.EclipseStarter.run(EclipseStarter.java:179) at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method) at sun.reflect.NativeMethodAccessorImpl.invoke(Unknown Source) at sun.reflect.DelegatingMethodAccessorImpl.invoke(Unknown Source) at java.lang.reflect.Method.invoke(Unknown Source) at org.eclipse.equinox.launcher.Main.invokeFramework(Main.java:559) at org.eclipse.equinox.launcher.Main.basicRun(Main.java:514) at org.eclipse.equinox.launcher.Main.run(Main.java:1311) Caused by: org.eclipse.core.commands.ExecutionException: Error exportingWar File at org.eclipse.jst.j2ee.internal.archive.operations.J2EEArtifactExportOperation.execute(J2EEArtifactExportOperation.java:131) at org.eclipse.wst.common.frameworks.internal.datamodel.DataModelPausibleOperationImpl$1.run(DataModelPausibleOperationImpl.java:376) at org.eclipse.core.internal.resources.Workspace.run(Workspace.java:1800) at org.eclipse.wst.common.frameworks.internal.datamodel.DataModelPausibleOperationImpl.runOperation(DataModelPausibleOperationImpl.java:401) at org.eclipse.wst.common.frameworks.internal.datamodel.DataModelPausibleOperationImpl.runOperation(DataModelPausibleOperationImpl.java:352) at org.eclipse.wst.common.frameworks.internal.datamodel.DataModelPausibleOperationImpl.doExecute(DataModelPausibleOperationImpl.java:242) at org.eclipse.wst.common.frameworks.internal.datamodel.DataModelPausibleOperationImpl.executeImpl(DataModelPausibleOperationImpl.java:214) at org.eclipse.wst.common.frameworks.internal.datamodel.DataModelPausibleOperationImpl.cacheThreadAndContinue(DataModelPausibleOperationImpl.java:89) at org.eclipse.wst.common.frameworks.internal.datamodel.DataModelPausibleOperationImpl.execute(DataModelPausibleOperationImpl.java:202) at org.eclipse.wst.common.frameworks.internal.datamodel.ui.DataModelWizard$1$CatchThrowableRunnableWithProgress.run(DataModelWizard.java:218) at org.eclipse.jface.operation.ModalContext$ModalContextThread.run(ModalContext.java:121) Caused by: org.eclipse.jst.j2ee.commonarchivecore.internal.exception.SaveFailureException: Error opening archive for export.. at org.eclipse.jst.j2ee.internal.web.archive.operations.WebComponentExportOperation.export(WebComponentExportOperation.java:64) at org.eclipse.jst.j2ee.internal.archive.operations.J2EEArtifactExportOperation.execute(J2EEArtifactExportOperation.java:123) ... 10 more Caused by: org.eclipse.jst.jee.archive.ArchiveSaveFailureException: Error saving archive: WebComponentArchiveLoadAdapter, Component: P/Nautilus2 to output path: D:/Nautilus2.war at org.eclipse.jst.jee.archive.internal.ArchiveFactoryImpl.saveArchive(ArchiveFactoryImpl.java:84) at org.eclipse.jst.j2ee.internal.archive.operations.J2EEArtifactExportOperation.saveArchive(J2EEArtifactExportOperation.java:306) at org.eclipse.jst.j2ee.internal.web.archive.operations.WebComponentExportOperation.export(WebComponentExportOperation.java:50) ... 11 more Caused by: java.io.FileNotFoundException: D:\myproject.war (Access is denied) at java.io.FileOutputStream.open(Native Method) at java.io.FileOutputStream.(Unknown Source) at java.io.FileOutputStream.(Unknown Source) at org.eclipse.jst.jee.archive.internal.ArchiveFactoryImpl.createSaveAdapterForJar(ArchiveFactoryImpl.java:108) at org.eclipse.jst.jee.archive.internal.ArchiveFactoryImpl.saveArchive(ArchiveFactoryImpl.java:74) ... 13 more Contains: Extended Operation failure: org.eclipse.jst.j2ee.internal.web.archive.operations.WebComponentExportOperation org.eclipse.core.commands.ExecutionException: Error exportingWar File at org.eclipse.jst.j2ee.internal.archive.operations.J2EEArtifactExportOperation.execute(J2EEArtifactExportOperation.java:131) at org.eclipse.wst.common.frameworks.internal.datamodel.DataModelPausibleOperationImpl$1.run(DataModelPausibleOperationImpl.java:376) at org.eclipse.core.internal.resources.Workspace.run(Workspace.java:1800) at org.eclipse.wst.common.frameworks.internal.datamodel.DataModelPausibleOperationImpl.runOperation(DataModelPausibleOperationImpl.java:401) at org.eclipse.wst.common.frameworks.internal.datamodel.DataModelPausibleOperationImpl.runOperation(DataModelPausibleOperationImpl.java:352) at org.eclipse.wst.common.frameworks.internal.datamodel.DataModelPausibleOperationImpl.doExecute(DataModelPausibleOperationImpl.java:242) at org.eclipse.wst.common.frameworks.internal.datamodel.DataModelPausibleOperationImpl.executeImpl(DataModelPausibleOperationImpl.java:214) at org.eclipse.wst.common.frameworks.internal.datamodel.DataModelPausibleOperationImpl.cacheThreadAndContinue(DataModelPausibleOperationImpl.java:89) at org.eclipse.wst.common.frameworks.internal.datamodel.DataModelPausibleOperationImpl.execute(DataModelPausibleOperationImpl.java:202) at org.eclipse.wst.common.frameworks.internal.datamodel.ui.DataModelWizard$1$CatchThrowableRunnableWithProgress.run(DataModelWizard.java:218) at org.eclipse.jface.operation.ModalContext$ModalContextThread.run(ModalContext.java:121) Caused by: org.eclipse.jst.j2ee.commonarchivecore.internal.exception.SaveFailureException: Error opening archive for export.. at org.eclipse.jst.j2ee.internal.web.archive.operations.WebComponentExportOperation.export(WebComponentExportOperation.java:64) at org.eclipse.jst.j2ee.internal.archive.operations.J2EEArtifactExportOperation.execute(J2EEArtifactExportOperation.java:123) ... 10 more Caused by: org.eclipse.jst.jee.archive.ArchiveSaveFailureException: Error saving archive: WebComponentArchiveLoadAdapter, Component: P/Nautilus2 to output path: D:/Nautilus2.war at org.eclipse.jst.jee.archive.internal.ArchiveFactoryImpl.saveArchive(ArchiveFactoryImpl.java:84) at org.eclipse.jst.j2ee.internal.archive.operations.J2EEArtifactExportOperation.saveArchive(J2EEArtifactExportOperation.java:306) at org.eclipse.jst.j2ee.internal.web.archive.operations.WebComponentExportOperation.export(WebComponentExportOperation.java:50) ... 11 more Caused by: java.io.FileNotFoundException: D:\myproject.war (Access is denied) at java.io.FileOutputStream.open(Native Method) at java.io.FileOutputStream.(Unknown Source) at java.io.FileOutputStream.(Unknown Source) at org.eclipse.jst.jee.archive.internal.ArchiveFactoryImpl.createSaveAdapterForJar(ArchiveFactoryImpl.java:108) at org.eclipse.jst.jee.archive.internal.ArchiveFactoryImpl.saveArchive(ArchiveFactoryImpl.java:74) ... 13 more Can anyone help me out with this ?

    Read the article

  • Avoiding the Anaemic Domain - How to decide what single responsibility a class has

    - by thecapsaicinkid
    Even after reading a bunch I'm still falling into the same trap. I have a class, usually an enity. I need to implement more than one, similar operations on this type. It feels wrong to (seemingly arbitrarily) choose one of these operations to belong inside the entity and push the others out to a separate class; I end up pushing all operations to service classes and am left with an anaemic domain. As a crude example, imagine the typical Employee class with numeric properties to hold how many paid days the employee is entitled to for both sickness and holiday and a collection of days taken for each. public class Employee { public int PaidHolidayAllowance { get; set; } public int PaidSicknessAllowance { get; set; } public IEnumerable<Holiday> Holidays { get; set; } public IEnumerable<SickDays> SickDays { get; set; } } I want two operations, one to calculate remaining holiday, another for remaining paid sick days. It seems strange to include say, CalculateRemaingHoliday() in the Employee class and bump CalculateRemainingPaidSick() to some PaidSicknessCalculator class. I would end up with a PaidSicknessCalculator and a RemainingHolidayCalculator and the anaemic Employee entity as seen above. The other alternative would be to put both operations in the Employee class and kick Single Responsibility to the curb. That doesn't make for particularly maintainable code. I suppose the Employee class should have some initialisation/validation logic (not accepting negative alowances etc.) So maybe I just stick to basic initialisation and validation in the entities themselves and be happy with my separate calculator classes. Or maybe I should be asking myself if Anaemic Domain is actually causing me some tangible problems with my code.

    Read the article

  • How can I keep directories in sync

    - by Guillaume Boudreau
    I have a directory, dirA, that users can work in: they can create, modify, rename and delete files & sub-directores in dirA. I want to keep another directory, dirB, in sync with dirA. What I'd like, is a discussion on finding a working algorithm that would achieve the above, with the limitations listed below. Requirements: 1. Something asynchronous - I don't want to stop file operations in dirA while I work in dirB. 2. I can't assume that I can just blindly rsync dirA to dirB on regular interval - dirA could contain millions of files & directories, and terrabytes of data. Completely walking the dirA tree could take hours. Those two requirements makes this really difficult. Having it asynchronous means that when I start working on a specific file from dirA, it might have moved a lot since it appeared. And the second limitation means that I really need to watch dirA, and work on atomic file operations that I notice. Current (broken) implementation: 1. Log all file & directory operations in dirA. 2. Using a separate process, read that log, and 'repeat' all the logged operations in dirB. Why is it broken: echo 1 > dirA/file1 # Allow the 'log reader' process to create dirB/file1: log = "write dirA/file1"; action = cp dirA/file1 dirB/file1; result = OK echo 1 > dirA/file2 mv dirA/file1 dirA/file3 mv dirA/file2 dirA/file1 rm dirA/file3 # End result: file1 contains '1' # 'log reader' process starts working on the 4 above file operations: log = "write file2"; action = cp dirA/file2 dirB/file2; result = failed: there is no dirA/file2 log = "rename file1 file3"; action = mv dirB/file1 dirB/file3; result = OK log = "rename file2 file1"; action = mv dirB/file2 dirB/file1; result = failed: there is no dirB/file2 log = "delete file3"; action = rm dirB/file3; result = OK # End result in dirB: no more files! Another broken example: echo 1 > dirA/dir1/file1 mv dirA/dir1 dirA/dir2 # 'log reader' process starts working on the 2 above file operations: log = "write file1"; action = cp dirA/dir1/file1 dirB/dir1/file1; result = failed: there is no dirA/dir1/file1 log = "rename dir1 dir2"; action = mv dirB/dir1 dirB/dir2; result = failed: there is no dirA/dir1 # End result if dirB: nothing!

    Read the article

  • How to Quickly Resize, Convert & Modify Images from the Linux Terminal

    - by Chris Hoffman
    ImageMagick is a suite of command-line utilities for modifying and working with images. ImageMagick can quickly perform operations on an image from a terminal, perform batch processing of many images, or be integrated into a bash script. ImageMagick can perform a wide variety of operations. This guide will introduce you to ImageMagick’s syntax and basic operations and show you how to combine operations and perform batch processing of many images. The HTG Guide to Hiding Your Data in a TrueCrypt Hidden Volume Make Your Own Windows 8 Start Button with Zero Memory Usage Reader Request: How To Repair Blurry Photos

    Read the article

  • What is Stackify?

    - by Matt Watson
    You have developers, applications, and servers. Stackify makes sure that they are all working efficiently. Our mission is to give developers the integrated tools they need to better troubleshoot and monitor the applications they create and the servers that they run on. Traditional IT operations tools are designed for network and system administrators. Developers commonly spend 30% of their time working with IT Operations remediating application service problems. Developers currently lack tools to efficiently support the applications they create. Stackify delivers the application support functionality that developers need:View application deployment locations, versions, and historyBrowse files on servers to ensure proper deploymentsAccess configuration and log files on serversRemotely restart windows services, scheduled tasks, and web applicationsBasic server monitoring and alertsCollects all application exceptions to a centralized pointLog and report on custom applications eventsStackify is building an integrated DevOps solution delivered from the cloud designed to meet the needs of developers but also help unify the working relationship with IT operations teams and existing security roles. Our goal is to help unify the interaction between developers and IT operations. Stackify allows both teams to have visibility that they never had before  to solve complex application service issues easier and faster. Stackify’s CEO and CTO both have experience managing very large and high growth software development teams. That experience is driving our design in Stackify to deliver the integrated tools we always wished we had, the next generation of development operations tools.

    Read the article

  • Online ALTER TABLE in MySQL 5.6

    - by Marko Mäkelä
    This is the low-level view of data dictionary language (DDL) operations in the InnoDB storage engine in MySQL 5.6. John Russell gave a more high-level view in his blog post April 2012 Labs Release – Online DDL Improvements. MySQL before the InnoDB Plugin Traditionally, the MySQL storage engine interface has taken a minimalistic approach to data definition language. The only natively supported operations were CREATE TABLE, DROP TABLE and RENAME TABLE. Consider the following example: CREATE TABLE t(a INT); INSERT INTO t VALUES (1),(2),(3); CREATE INDEX a ON t(a); DROP TABLE t; The CREATE INDEX statement would be executed roughly as follows: CREATE TABLE temp(a INT, INDEX(a)); INSERT INTO temp SELECT * FROM t; RENAME TABLE t TO temp2; RENAME TABLE temp TO t; DROP TABLE temp2; You could imagine that the database could crash when copying all rows from the original table to the new one. For example, it could run out of file space. Then, on restart, InnoDB would roll back the huge INSERT transaction. To fix things a little, a hack was added to ha_innobase::write_row for committing the transaction every 10,000 rows. Still, it was frustrating that even a simple DROP INDEX would make the table unavailable for modifications for a long time. Fast Index Creation in the InnoDB Plugin of MySQL 5.1 MySQL 5.1 introduced a new interface for CREATE INDEX and DROP INDEX. The old table-copying approach can still be forced by SET old_alter_table=0. This interface is used in MySQL 5.5 and in the InnoDB Plugin for MySQL 5.1. Apart from the ability to do a quick DROP INDEX, the main advantage is that InnoDB will execute a merge-sort algorithm before inserting the index records into each index that is being created. This should speed up the insert into the secondary index B-trees and potentially result in a better B-tree fill factor. The 5.1 ALTER TABLE interface was not perfect. For example, DROP FOREIGN KEY still invoked the table copy. Renaming columns could conflict with InnoDB foreign key constraints. Combining ADD KEY and DROP KEY in ALTER TABLE was problematic and not atomic inside the storage engine. The ALTER TABLE interface in MySQL 5.6 The ALTER TABLE storage engine interface was completely rewritten in MySQL 5.6. Instead of introducing a method call for every conceivable operation, MySQL 5.6 introduced a handful of methods, and data structures that keep track of the requested changes. In MySQL 5.6, online ALTER TABLE operation can be requested by specifying LOCK=NONE. Also LOCK=SHARED and LOCK=EXCLUSIVE are available. The old-style table copying can be requested by ALGORITHM=COPY. That one will require at least LOCK=SHARED. From the InnoDB point of view, anything that is possible with LOCK=EXCLUSIVE is also possible with LOCK=SHARED. Most ALGORITHM=INPLACE operations inside InnoDB can be executed online (LOCK=NONE). InnoDB will always require an exclusive table lock in two phases of the operation. The execution phases are tied to a number of methods: handler::check_if_supported_inplace_alter Checks if the storage engine can perform all requested operations, and if so, what kind of locking is needed. handler::prepare_inplace_alter_table InnoDB uses this method to set up the data dictionary cache for upcoming CREATE INDEX operation. We need stubs for the new indexes, so that we can keep track of changes to the table during online index creation. Also, crash recovery would drop any indexes that were incomplete at the time of the crash. handler::inplace_alter_table In InnoDB, this method is used for creating secondary indexes or for rebuilding the table. This is the ‘main’ phase that can be executed online (with concurrent writes to the table). handler::commit_inplace_alter_table This is where the operation is committed or rolled back. Here, InnoDB would drop any indexes, rename any columns, drop or add foreign keys, and finalize a table rebuild or index creation. It would also discard any logs that were set up for online index creation or table rebuild. The prepare and commit phases require an exclusive lock, blocking all access to the table. If MySQL times out while upgrading the table meta-data lock for the commit phase, it will roll back the ALTER TABLE operation. In MySQL 5.6, data definition language operations are still not fully atomic, because the data dictionary is split. Part of it is inside InnoDB data dictionary tables. Part of the information is only available in the *.frm file, which is not covered by any crash recovery log. But, there is a single commit phase inside the storage engine. Online Secondary Index Creation It may occur that an index needs to be created on a new column to speed up queries. But, it may be unacceptable to block modifications on the table while creating the index. It turns out that it is conceptually not so hard to support online index creation. All we need is some more execution phases: Set up a stub for the index, for logging changes. Scan the table for index records. Sort the index records. Bulk load the index records. Apply the logged changes. Replace the stub with the actual index. Threads that modify the table will log the operations to the logs of each index that is being created. Errors, such as log overflow or uniqueness violations, will only be flagged by the ALTER TABLE thread. The log is conceptually similar to the InnoDB change buffer. The bulk load of index records will bypass record locking. We still generate redo log for writing the index pages. It would suffice to log page allocations only, and to flush the index pages from the buffer pool to the file system upon completion. Native ALTER TABLE Starting with MySQL 5.6, InnoDB supports most ALTER TABLE operations natively. The notable exceptions are changes to the column type, ADD FOREIGN KEY except when foreign_key_checks=0, and changes to tables that contain FULLTEXT indexes. The keyword ALGORITHM=INPLACE is somewhat misleading, because certain operations cannot be performed in-place. For example, changing the ROW_FORMAT of a table requires a rebuild. Online operation (LOCK=NONE) is not allowed in the following cases: when adding an AUTO_INCREMENT column, when the table contains FULLTEXT indexes or a hidden FTS_DOC_ID column, or when there are FOREIGN KEY constraints referring to the table, with ON…CASCADE or ON…SET NULL option. The FOREIGN KEY limitations are needed, because MySQL does not acquire meta-data locks on the child or parent tables when executing SQL statements. Theoretically, InnoDB could support operations like ADD COLUMN and DROP COLUMN in-place, by lazily converting the table to a newer format. This would require that the data dictionary keep multiple versions of the table definition. For simplicity, we will copy the entire table, even for DROP COLUMN. The bulk copying of the table will bypass record locking and undo logging. For facilitating online operation, a temporary log will be associated with the clustered index of table. Threads that modify the table will also write the changes to the log. When altering the table, we skip all records that have been marked for deletion. In this way, we can simply discard any undo log records that were not yet purged from the original table. Off-page columns, or BLOBs, are an important consideration. We suspend the purge of delete-marked records if it would free any off-page columns from the old table. This is because the BLOBs can be needed when applying changes from the log. We have special logging for handling the ROLLBACK of an INSERT that inserted new off-page columns. This is because the columns will be freed at rollback.

    Read the article

  • ARTS Reference Model for Retail

    - by Sanjeev Sharma
    Consider a hypothetical scenario where you have been tasked to set up retail operations for a electronic goods or daily consumables or a luxury brand etc. It is very likely you will be faced with the following questions: What are the essential business capabilities that you must have in place?  What are the essential business activities under-pinning each of the business capabilities, identified in Step 1? What are the set of steps that you need to perform to execute each of the business activities, identified in Step 2? Answers to the above will drive your investments in software and hardware to enable the core retail operations. More importantly, the choices you make in responding to the above questions will several implications in the short-run and in the long-run. In the short-term, you will incur the time and cost of defining your technology requirements, procuring the software/hardware components and getting them up and running. In the long-term, as you grow in operations organically or through M&A, partnerships and franchiser business models  you will invariably need to make more technology investments to manage the greater complexity (scale and scope) of business operations.  "As new software applications, such as time & attendance, labor scheduling, and POS transactions, just to mention a few, are introduced into the store environment, it takes a disproportionate amount of time and effort to integrate them with existing store applications. These integration projects can add up to 50 percent to the time needed to implement a new software application and contribute significantly to the cost of the overall project, particularly if a systems integrator is called in. This has been the reality that all retailers have had to live with over the last two decades. The effect of the environment has not only been to increase costs, but also to limit retailers' ability to implement change and the speed with which they can do so." (excerpt taken from here) Now, one would think a lot of retailers would have already gone through the pain of finding answers to these questions, so why re-invent the wheel? Precisely so, a major effort began almost 17 years ago in the retail industry to make it less expensive and less difficult to deploy new technology in stores and at the retail enterprise level. This effort is called the Association for Retail Technology Standards (ARTS). Without standards such as those defined by ARTS, you would very likely end up experiencing the following: Increased Time and Cost due to resource wastage arising from re-inventing the wheel i.e. re-creating vanilla processes from scratch, and incurring, otherwise avoidable, mistakes and errors by ignoring experience of others Sub-optimal Process Efficiency due to narrow, isolated view of processes thereby ignoring process inter-dependencies i.e. optimizing parts but not the whole, and resulting in lack of transparency and inter-departmental finger-pointing Embracing ARTS standards as a blue-print for establishing or managing or streamlining your retail operations can benefit you in the following ways: Improved Time-to-Market from parity with industry best-practice processes e.g. ARTS, thus avoiding “reinventing the wheel” for common retail processes and focusing more on customizing processes for differentiations, and lowering integration complexity and risk with a standardized vocabulary for exchange between internal and external i.e. partner systems Lower Operating Costs by embracing the ARTS enterprise-wide process reference model for developing and streamlining retail operations holistically instead of a narrow, silo-ed view, and  procuring IT systems in compliance with ARTS thus avoiding IT budget marginalization While parity with industry standards such as ARTS business process model by itself does not create a differentiation, it does however provide a higher starting point for bridging the strategy-execution gap in setting up and improving retail operations.

    Read the article

  • Flags with deferred use

    - by Trenton Maki
    Let's say I have a system. In this system I have a number of operations I can do but all of these operations have to happen as a batch at a certain time, while calls to activate and deactivate these operations can come in at any time. To implement this, I could use flags like doOperation1 and doOperation2 but this seems like it would become difficult to maintain. Is there a design pattern, or something similar, that addresses this situation?

    Read the article

  • Multiple vulnerabilities in libexif

    - by Umang_D
    CVE DescriptionCVSSv2 Base ScoreComponentProduct and Resolution CVE-2012-2812 Improper Restriction of Operations within the Bounds of a Memory Buffer vulnerability 6.4 libexif Solaris 11 11/11 SRU 12.4 CVE-2012-2813 Improper Restriction of Operations within the Bounds of a Memory Buffer vulnerability 6.4 CVE-2012-2814 Improper Restriction of Operations within the Bounds of a Memory Buffer vulnerability 7.5 CVE-2012-2836 Improper Restriction of Operations within the Bounds of a Memory Buffer vulnerability 6.4 CVE-2012-2837 Numeric Errors vulnerability 5.0 CVE-2012-2840 Numeric Errors vulnerability 7.5 CVE-2012-2841 Numeric Errors vulnerability 7.5 CVE-2012-2845 Numeric Errors vulnerability 6.4 This notification describes vulnerabilities fixed in third-party components that are included in Oracle's product distributions.Information about vulnerabilities affecting Oracle products can be found on Oracle Critical Patch Updates and Security Alerts page.

    Read the article

  • Maximizing Throughput with TVPs

    TVPs offer several performance optimization possibilities that other bulk operations do not allow, and these operations may allow for TVP performance to exceed other bulk operations by an order of magnitude, especially for a pattern where subsets of the data are frequently updated. Want to work faster with SQL Server?If you want to work faster try out the SQL Toolbelt. "The SQL Toolbelt provides tools that database developers as well as DBAs should not live without." William Van Orden. Download the SQL Toolbelt here.

    Read the article

  • How to make write operation idempotent?

    - by Morgan Cheng
    I'm reading article about recently release Gizzard sharding framework by twitter(http://engineering.twitter.com/2010/04/introducing-gizzard-framework-for.html). It mentions that all write operations must be idempotent to make sure high reliability. According to wikipedia, "Idempotent operations are operations that can be applied multiple times without changing the result." But, IMHO, in Gazzard case, idempotent write operation should be operations that sequence doesn't matter. Now, my question is: How to make write operation idempotent? The only thing I can image is to have a version number attached to each write. For example, in blog system. Each blog must have a $blog_id and $content. In application level, we always write a blog content like this write($blog_id, $content, $version). The $version is determined to be unique in application level. So, if application first try to set one blog to "Hello world" and second want it to be "Goodbye", the write is idempotent. We have such two write operations: write($blog_id, "Hello world", 1); write($blog_id, "Goodbye", 2); These two operations are supposed to changed two different records in DB. So, no matter how many times and what sequence these two operations executed, the results are same. This is just my understanding. Please correct me if I'm wrong.

    Read the article

  • Old operational master still thinks it is the "one"

    - by Doug
    Hi there, I have a domain with 3 AD servers for now i'll just call them: AD01 (Win 2008 GC, Operations master) AD02 (Win 2008 GC) AD03 (Win 2003 GC) A couple of months there was some hardware issues with AD01 so the operations master, PDC and Infrastructure Master was moved to AD02. All machines where on while this was happening. AD01 (Win 2008 GC) AD02 (Win 2008 GC, Operations master) AD03 (Win 2003 GC) AD01 was then shutdown for a month. Upon starting this machine up with replaced hardware (NIC and RAID card) i now have a weird problem. AD01 Thinks it is operations master still in AD on the local box AD02 & AD03 Thinks AD02 is operations master in AD on both boxes When running DCDIAG on AD01 i get a number of issues (listed below) When running "dcdiag /test:advertising" on AD01: Doing primary tests Testing server: Default-First-Site-Name\AD01 Starting test: Advertising Warning: DsGetDcName returned information for \\ad02.domain.local, when we were trying to reach AD01. SERVER IS NOT RESPONDING or IS NOT CONSIDERED SUITABLE. ......................... AD01 failed test Advertising Running partition tests on : ForestDnsZones Running partition tests on : DomainDnsZones Running partition tests on : Schema Running partition tests on : Configuration Running partition tests on : domain Running enterprise tests on : domain.local When running "dcdiag" on AD01 i get the following errors (excerpt of the Final output): Testing server: Default-First-Site-Name\AD01 Starting test: Advertising Warning: DsGetDcName returned information for \\ad02.domain.local, when we were trying to reach AD01. SERVER IS NOT RESPONDING or IS NOT CONSIDERED SUITABLE. ......................... AD01 failed test Advertising Starting test: FrsEvent There are warning or error events within the last 24 hours after the SYSVOL has been shared. Failing SYSVOL replication problems may cause Group Policy problems. Starting test: NCSecDesc Error NT AUTHORITY\ENTERPRISE DOMAIN CONTROLLERS doesn't have Replicating Directory Changes In Filtered Set access rights for the naming context: DC=ForestDnsZones,DC=domain,DC=local Error NT AUTHORITY\ENTERPRISE DOMAIN CONTROLLERS doesn't have Replicating Directory Changes In Filtered Set access rights for the naming context: DC=DomainDnsZones,DC=domain,DC=local Starting test: Replications [Replications Check,Replications Check] Inbound replication is disabled. To correct, run "repadmin /options AD01 -DISABLE_INBOUND_REPL" [Replications Check,AD01] Outbound replication is disabled. To correct, run "repadmin /options AD01 -DISABLE_OUTBOUND_REPL" So the problem appeasr to be that when i moved the operations master, AD01 never got the memo, and now that it's started up, all the other AD servers don't think its the boss anymore when it trys to replicate etc. So i really need to manually update AD01 so that it knows who the operations master, instrastructure and PDC is - but i'm not having any luck I've been googling for nearly a day and all solutions lead to "the cake is a lie" Your ninja skills will be greatly appreciated

    Read the article

  • C#/.NET Little Wonders: The Useful But Overlooked Sets

    - by James Michael Hare
    Once again we consider some of the lesser known classes and keywords of C#.  Today we will be looking at two set implementations in the System.Collections.Generic namespace: HashSet<T> and SortedSet<T>.  Even though most people think of sets as mathematical constructs, they are actually very useful classes that can be used to help make your application more performant if used appropriately. A Background From Math In mathematical terms, a set is an unordered collection of unique items.  In other words, the set {2,3,5} is identical to the set {3,5,2}.  In addition, the set {2, 2, 4, 1} would be invalid because it would have a duplicate item (2).  In addition, you can perform set arithmetic on sets such as: Intersections: The intersection of two sets is the collection of elements common to both.  Example: The intersection of {1,2,5} and {2,4,9} is the set {2}. Unions: The union of two sets is the collection of unique items present in either or both set.  Example: The union of {1,2,5} and {2,4,9} is {1,2,4,5,9}. Differences: The difference of two sets is the removal of all items from the first set that are common between the sets.  Example: The difference of {1,2,5} and {2,4,9} is {1,5}. Supersets: One set is a superset of a second set if it contains all elements that are in the second set. Example: The set {1,2,5} is a superset of {1,5}. Subsets: One set is a subset of a second set if all the elements of that set are contained in the first set. Example: The set {1,5} is a subset of {1,2,5}. If We’re Not Doing Math, Why Do We Care? Now, you may be thinking: why bother with the set classes in C# if you have no need for mathematical set manipulation?  The answer is simple: they are extremely efficient ways to determine ownership in a collection. For example, let’s say you are designing an order system that tracks the price of a particular equity, and once it reaches a certain point will trigger an order.  Now, since there’s tens of thousands of equities on the markets, you don’t want to track market data for every ticker as that would be a waste of time and processing power for symbols you don’t have orders for.  Thus, we just want to subscribe to the stock symbol for an equity order only if it is a symbol we are not already subscribed to. Every time a new order comes in, we will check the list of subscriptions to see if the new order’s stock symbol is in that list.  If it is, great, we already have that market data feed!  If not, then and only then should we subscribe to the feed for that symbol. So far so good, we have a collection of symbols and we want to see if a symbol is present in that collection and if not, add it.  This really is the essence of set processing, but for the sake of comparison, let’s say you do a list instead: 1: // class that handles are order processing service 2: public sealed class OrderProcessor 3: { 4: // contains list of all symbols we are currently subscribed to 5: private readonly List<string> _subscriptions = new List<string>(); 6:  7: ... 8: } Now whenever you are adding a new order, it would look something like: 1: public PlaceOrderResponse PlaceOrder(Order newOrder) 2: { 3: // do some validation, of course... 4:  5: // check to see if already subscribed, if not add a subscription 6: if (!_subscriptions.Contains(newOrder.Symbol)) 7: { 8: // add the symbol to the list 9: _subscriptions.Add(newOrder.Symbol); 10: 11: // do whatever magic is needed to start a subscription for the symbol 12: } 13:  14: // place the order logic! 15: } What’s wrong with this?  In short: performance!  Finding an item inside a List<T> is a linear - O(n) – operation, which is not a very performant way to find if an item exists in a collection. (I used to teach algorithms and data structures in my spare time at a local university, and when you began talking about big-O notation you could immediately begin to see eyes glossing over as if it was pure, useless theory that would not apply in the real world, but I did and still do believe it is something worth understanding well to make the best choices in computer science). Let’s think about this: a linear operation means that as the number of items increases, the time that it takes to perform the operation tends to increase in a linear fashion.  Put crudely, this means if you double the collection size, you might expect the operation to take something like the order of twice as long.  Linear operations tend to be bad for performance because they mean that to perform some operation on a collection, you must potentially “visit” every item in the collection.  Consider finding an item in a List<T>: if you want to see if the list has an item, you must potentially check every item in the list before you find it or determine it’s not found. Now, we could of course sort our list and then perform a binary search on it, but sorting is typically a linear-logarithmic complexity – O(n * log n) - and could involve temporary storage.  So performing a sort after each add would probably add more time.  As an alternative, we could use a SortedList<TKey, TValue> which sorts the list on every Add(), but this has a similar level of complexity to move the items and also requires a key and value, and in our case the key is the value. This is why sets tend to be the best choice for this type of processing: they don’t rely on separate keys and values for ordering – so they save space – and they typically don’t care about ordering – so they tend to be extremely performant.  The .NET BCL (Base Class Library) has had the HashSet<T> since .NET 3.5, but at that time it did not implement the ISet<T> interface.  As of .NET 4.0, HashSet<T> implements ISet<T> and a new set, the SortedSet<T> was added that gives you a set with ordering. HashSet<T> – For Unordered Storage of Sets When used right, HashSet<T> is a beautiful collection, you can think of it as a simplified Dictionary<T,T>.  That is, a Dictionary where the TKey and TValue refer to the same object.  This is really an oversimplification, but logically it makes sense.  I’ve actually seen people code a Dictionary<T,T> where they store the same thing in the key and the value, and that’s just inefficient because of the extra storage to hold both the key and the value. As it’s name implies, the HashSet<T> uses a hashing algorithm to find the items in the set, which means it does take up some additional space, but it has lightning fast lookups!  Compare the times below between HashSet<T> and List<T>: Operation HashSet<T> List<T> Add() O(1) O(1) at end O(n) in middle Remove() O(1) O(n) Contains() O(1) O(n)   Now, these times are amortized and represent the typical case.  In the very worst case, the operations could be linear if they involve a resizing of the collection – but this is true for both the List and HashSet so that’s a less of an issue when comparing the two. The key thing to note is that in the general case, HashSet is constant time for adds, removes, and contains!  This means that no matter how large the collection is, it takes roughly the exact same amount of time to find an item or determine if it’s not in the collection.  Compare this to the List where almost any add or remove must rearrange potentially all the elements!  And to find an item in the list (if unsorted) you must search every item in the List. So as you can see, if you want to create an unordered collection and have very fast lookup and manipulation, the HashSet is a great collection. And since HashSet<T> implements ICollection<T> and IEnumerable<T>, it supports nearly all the same basic operations as the List<T> and can use the System.Linq extension methods as well. All we have to do to switch from a List<T> to a HashSet<T>  is change our declaration.  Since List and HashSet support many of the same members, chances are we won’t need to change much else. 1: public sealed class OrderProcessor 2: { 3: private readonly HashSet<string> _subscriptions = new HashSet<string>(); 4:  5: // ... 6:  7: public PlaceOrderResponse PlaceOrder(Order newOrder) 8: { 9: // do some validation, of course... 10: 11: // check to see if already subscribed, if not add a subscription 12: if (!_subscriptions.Contains(newOrder.Symbol)) 13: { 14: // add the symbol to the list 15: _subscriptions.Add(newOrder.Symbol); 16: 17: // do whatever magic is needed to start a subscription for the symbol 18: } 19: 20: // place the order logic! 21: } 22:  23: // ... 24: } 25: Notice, we didn’t change any code other than the declaration for _subscriptions to be a HashSet<T>.  Thus, we can pick up the performance improvements in this case with minimal code changes. SortedSet<T> – Ordered Storage of Sets Just like HashSet<T> is logically similar to Dictionary<T,T>, the SortedSet<T> is logically similar to the SortedDictionary<T,T>. The SortedSet can be used when you want to do set operations on a collection, but you want to maintain that collection in sorted order.  Now, this is not necessarily mathematically relevant, but if your collection needs do include order, this is the set to use. So the SortedSet seems to be implemented as a binary tree (possibly a red-black tree) internally.  Since binary trees are dynamic structures and non-contiguous (unlike List and SortedList) this means that inserts and deletes do not involve rearranging elements, or changing the linking of the nodes.  There is some overhead in keeping the nodes in order, but it is much smaller than a contiguous storage collection like a List<T>.  Let’s compare the three: Operation HashSet<T> SortedSet<T> List<T> Add() O(1) O(log n) O(1) at end O(n) in middle Remove() O(1) O(log n) O(n) Contains() O(1) O(log n) O(n)   The MSDN documentation seems to indicate that operations on SortedSet are O(1), but this seems to be inconsistent with its implementation and seems to be a documentation error.  There’s actually a separate MSDN document (here) on SortedSet that indicates that it is, in fact, logarithmic in complexity.  Let’s put it in layman’s terms: logarithmic means you can double the collection size and typically you only add a single extra “visit” to an item in the collection.  Take that in contrast to List<T>’s linear operation where if you double the size of the collection you double the “visits” to items in the collection.  This is very good performance!  It’s still not as performant as HashSet<T> where it always just visits one item (amortized), but for the addition of sorting this is a good thing. Consider the following table, now this is just illustrative data of the relative complexities, but it’s enough to get the point: Collection Size O(1) Visits O(log n) Visits O(n) Visits 1 1 1 1 10 1 4 10 100 1 7 100 1000 1 10 1000   Notice that the logarithmic – O(log n) – visit count goes up very slowly compare to the linear – O(n) – visit count.  This is because since the list is sorted, it can do one check in the middle of the list, determine which half of the collection the data is in, and discard the other half (binary search).  So, if you need your set to be sorted, you can use the SortedSet<T> just like the HashSet<T> and gain sorting for a small performance hit, but it’s still faster than a List<T>. Unique Set Operations Now, if you do want to perform more set-like operations, both implementations of ISet<T> support the following, which play back towards the mathematical set operations described before: IntersectWith() – Performs the set intersection of two sets.  Modifies the current set so that it only contains elements also in the second set. UnionWith() – Performs a set union of two sets.  Modifies the current set so it contains all elements present both in the current set and the second set. ExceptWith() – Performs a set difference of two sets.  Modifies the current set so that it removes all elements present in the second set. IsSupersetOf() – Checks if the current set is a superset of the second set. IsSubsetOf() – Checks if the current set is a subset of the second set. For more information on the set operations themselves, see the MSDN description of ISet<T> (here). What Sets Don’t Do Don’t get me wrong, sets are not silver bullets.  You don’t really want to use a set when you want separate key to value lookups, that’s what the IDictionary implementations are best for. Also sets don’t store temporal add-order.  That is, if you are adding items to the end of a list all the time, your list is ordered in terms of when items were added to it.  This is something the sets don’t do naturally (though you could use a SortedSet with an IComparer with a DateTime but that’s overkill) but List<T> can. Also, List<T> allows indexing which is a blazingly fast way to iterate through items in the collection.  Iterating over all the items in a List<T> is generally much, much faster than iterating over a set. Summary Sets are an excellent tool for maintaining a lookup table where the item is both the key and the value.  In addition, if you have need for the mathematical set operations, the C# sets support those as well.  The HashSet<T> is the set of choice if you want the fastest possible lookups but don’t care about order.  In contrast the SortedSet<T> will give you a sorted collection at a slight reduction in performance.   Technorati Tags: C#,.Net,Little Wonders,BlackRabbitCoder,ISet,HashSet,SortedSet

    Read the article

  • CRUD operations; do you notify whether the insert,update etc. went well ?

    - by danielovich
    Hi guys. I have a simple question for you (i hope) :) I have pretty much always used void as a "return" type when doing CRUD operations on data. Eg. Consider this code: public void Insert(IAuctionItem item) { if (item == null) { AuctionLogger.LogException(new ArgumentNullException("item is null")); } _dataStore.DataContext.AuctionItems.InsertOnSubmit((AuctionItem)item); _dataStore.DataContext.SubmitChanges(); } and then considen this code: public bool Insert(IAuctionItem item) { if (item == null) { AuctionLogger.LogException(new ArgumentNullException("item is null")); } _dataStore.DataContext.AuctionItems.InsertOnSubmit((AuctionItem)item); _dataStore.DataContext.SubmitChanges(); return true; } It actually just comes down to whether you should notify that something was inserted (and went well) or not ?

    Read the article

  • Transactional Messaging in the Windows Azure Service Bus

    - by Alan Smith
    Introduction I’m currently working on broadening the content in the Windows Azure Service Bus Developer Guide. One of the features I have been looking at over the past week is the support for transactional messaging. When using the direct programming model and the WCF interface some, but not all, messaging operations can participate in transactions. This allows developers to improve the reliability of messaging systems. There are some limitations in the transactional model, transactions can only include one top level messaging entity (such as a queue or topic, subscriptions are no top level entities), and transactions cannot include other systems, such as databases. As the transaction model is currently not well documented I have had to figure out how things work through experimentation, with some help from the development team to confirm any questions I had. Hopefully I’ve got the content mostly correct, I will update the content in the e-book if I find any errors or improvements that can be made (any feedback would be very welcome). I’ve not had a chance to look into the code for transactions and asynchronous operations, maybe that would make a nice challenge lab for my Windows Azure Service Bus course. Transactional Messaging Messaging entities in the Windows Azure Service Bus provide support for participation in transactions. This allows developers to perform several messaging operations within a transactional scope, and ensure that all the actions are committed or, if there is a failure, none of the actions are committed. There are a number of scenarios where the use of transactions can increase the reliability of messaging systems. Using TransactionScope In .NET the TransactionScope class can be used to perform a series of actions in a transaction. The using declaration is typically used de define the scope of the transaction. Any transactional operations that are contained within the scope can be committed by calling the Complete method. If the Complete method is not called, any transactional methods in the scope will not commit.   // Create a transactional scope. using (TransactionScope scope = new TransactionScope()) {     // Do something.       // Do something else.       // Commit the transaction.     scope.Complete(); }     In order for methods to participate in the transaction, they must provide support for transactional operations. Database and message queue operations typically provide support for transactions. Transactions in Brokered Messaging Transaction support in Service Bus Brokered Messaging allows message operations to be performed within a transactional scope; however there are some limitations around what operations can be performed within the transaction. In the current release, only one top level messaging entity, such as a queue or topic can participate in a transaction, and the transaction cannot include any other transaction resource managers, making transactions spanning a messaging entity and a database not possible. When sending messages, the send operations can participate in a transaction allowing multiple messages to be sent within a transactional scope. This allows for “all or nothing” delivery of a series of messages to a single queue or topic. When receiving messages, messages that are received in the peek-lock receive mode can be completed, deadlettered or deferred within a transactional scope. In the current release the Abandon method will not participate in a transaction. The same restrictions of only one top level messaging entity applies here, so the Complete method can be called transitionally on messages received from the same queue, or messages received from one or more subscriptions in the same topic. Sending Multiple Messages in a Transaction A transactional scope can be used to send multiple messages to a queue or topic. This will ensure that all the messages will be enqueued or, if the transaction fails to commit, no messages will be enqueued.     An example of the code used to send 10 messages to a queue as a single transaction from a console application is shown below.   QueueClient queueClient = messagingFactory.CreateQueueClient(Queue1);   Console.Write("Sending");   // Create a transaction scope. using (TransactionScope scope = new TransactionScope()) {     for (int i = 0; i < 10; i++)     {         // Send a message         BrokeredMessage msg = new BrokeredMessage("Message: " + i);         queueClient.Send(msg);         Console.Write(".");     }     Console.WriteLine("Done!");     Console.WriteLine();       // Should we commit the transaction?     Console.WriteLine("Commit send 10 messages? (yes or no)");     string reply = Console.ReadLine();     if (reply.ToLower().Equals("yes"))     {         // Commit the transaction.         scope.Complete();     } } Console.WriteLine(); messagingFactory.Close();     The transaction scope is used to wrap the sending of 10 messages. Once the messages have been sent the user has the option to either commit the transaction or abandon the transaction. If the user enters “yes”, the Complete method is called on the scope, which will commit the transaction and result in the messages being enqueued. If the user enters anything other than “yes”, the transaction will not commit, and the messages will not be enqueued. Receiving Multiple Messages in a Transaction The receiving of multiple messages is another scenario where the use of transactions can improve reliability. When receiving a group of messages that are related together, maybe in the same message session, it is possible to receive the messages in the peek-lock receive mode, and then complete, defer, or deadletter the messages in one transaction. (In the current version of Service Bus, abandon is not transactional.)   The following code shows how this can be achieved. using (TransactionScope scope = new TransactionScope()) {       while (true)     {         // Receive a message.         BrokeredMessage msg = q1Client.Receive(TimeSpan.FromSeconds(1));         if (msg != null)         {             // Wrote message body and complete message.             string text = msg.GetBody<string>();             Console.WriteLine("Received: " + text);             msg.Complete();         }         else         {             break;         }     }     Console.WriteLine();       // Should we commit?     Console.WriteLine("Commit receive? (yes or no)");     string reply = Console.ReadLine();     if (reply.ToLower().Equals("yes"))     {         // Commit the transaction.         scope.Complete();     }     Console.WriteLine(); }     Note that if there are a large number of messages to be received, there will be a chance that the transaction may time out before it can be committed. It is possible to specify a longer timeout when the transaction is created, but It may be better to receive and commit smaller amounts of messages within the transaction. It is also possible to complete, defer, or deadletter messages received from more than one subscription, as long as all the subscriptions are contained in the same topic. As subscriptions are not top level messaging entities this scenarios will work. The following code shows how this can be achieved. try {     using (TransactionScope scope = new TransactionScope())     {         // Receive one message from each subscription.         BrokeredMessage msg1 = subscriptionClient1.Receive();         BrokeredMessage msg2 = subscriptionClient2.Receive();           // Complete the message receives.         msg1.Complete();         msg2.Complete();           Console.WriteLine("Msg1: " + msg1.GetBody<string>());         Console.WriteLine("Msg2: " + msg2.GetBody<string>());           // Commit the transaction.         scope.Complete();     } } catch (Exception ex) {     Console.WriteLine(ex.Message); }     Unsupported Scenarios The restriction of only one top level messaging entity being able to participate in a transaction makes some useful scenarios unsupported. As the Windows Azure Service Bus is under continuous development and new releases are expected to be frequent it is possible that this restriction may not be present in future releases. The first is the scenario where messages are to be routed to two different systems. The following code attempts to do this.   try {     // Create a transaction scope.     using (TransactionScope scope = new TransactionScope())     {         BrokeredMessage msg1 = new BrokeredMessage("Message1");         BrokeredMessage msg2 = new BrokeredMessage("Message2");           // Send a message to Queue1         Console.WriteLine("Sending Message1");         queue1Client.Send(msg1);           // Send a message to Queue2         Console.WriteLine("Sending Message2");         queue2Client.Send(msg2);           // Commit the transaction.         Console.WriteLine("Committing transaction...");         scope.Complete();     } } catch (Exception ex) {     Console.WriteLine(ex.Message); }     The results of running the code are shown below. When attempting to send a message to the second queue the following exception is thrown: No active Transaction was found for ID '35ad2495-ee8a-4956-bbad-eb4fedf4a96e:1'. The Transaction may have timed out or attempted to span multiple top-level entities such as Queue or Topic. The server Transaction timeout is: 00:01:00..TrackingId:947b8c4b-7754-4044-b91b-4a959c3f9192_3_3,TimeStamp:3/29/2012 7:47:32 AM.   Another scenario where transactional support could be useful is when forwarding messages from one queue to another queue. This would also involve more than one top level messaging entity, and is therefore not supported.   Another scenario that developers may wish to implement is performing transactions across messaging entities and other transactional systems, such as an on-premise database. In the current release this is not supported.   Workarounds for Unsupported Scenarios There are some techniques that developers can use to work around the one top level entity limitation of transactions. When sending two messages to two systems, topics and subscriptions can be used. If the same message is to be sent to two destinations then the subscriptions would have the default subscriptions, and the client would only send one message. If two different messages are to be sent, then filters on the subscriptions can route the messages to the appropriate destination. The client can then send the two messages to the topic in the same transaction.   In scenarios where a message needs to be received and then forwarded to another system within the same transaction topics and subscriptions can also be used. A message can be received from a subscription, and then sent to a topic within the same transaction. As a topic is a top level messaging entity, and a subscription is not, this scenario will work.

    Read the article

  • Agile Development Requires Agile Support

    - by Matt Watson
    Agile developmentAgile development has become the standard methodology for application development. The days of long term planning with giant Gantt waterfall charts and detailed requirements is fading away. For years the product planning process frustrated product owners and businesses because no matter the plan, nothing ever went to plan. Agile development throws the detailed planning out the window and instead focuses on giving developers some basic requirements and pointing them in the right direction. Constant collaboration via quick iterations with the end users, product owners, and the development team helps ensure the project is done correctly.  The various agile development methodologies have helped greatly with creating products faster, but not without causing new problems. Complicated application deployments now occur weekly or monthly. Most of the products are web-based and deployed as a software service model. System performance and availability of these apps becomes mission critical. This is all much different from the old process of mailing new releases of client-server apps on CD once per quarter or year.The steady stream of new products and product enhancements puts a lot of pressure on IT operations to keep up with the software deployments and adding infrastructure capacity. The problem is most operations teams still move slowly thanks to change orders, documentation, procedures, testing and other processes. Operations can slow the process down and push back on the development team in some organizations. The DevOps movement is trying to solve some of these problems by integrating the development and operations teams more together. Rapid change introduces new problemsThe rapid product change ultimately creates some application problems along the way. Higher rates of change increase the likelihood of new application defects. Delivering applications as a software service also means that scalability of applications is critical. Development teams struggle to keep up with application defects and scalability concerns in their applications. Fixing application problems is a never ending job for agile development teams. Fixing problems before your customers do and fixing them quickly is critical. Most companies really struggle with this due to the divide between the development and operations groups. Fixing application problems typically requires querying databases, looking at log files, reviewing config files, reviewing error logs and other similar tasks. It becomes difficult to work on new features when your lead developers are working on defects from the last product version. Developers need more visibilityThe problem is most developers are not given access to see server and application information in the production environments. The operations team doesn’t trust giving all the developers the keys to the kingdom to log in to production and poke around the servers. The challenge is either give them no access, or potentially too much access. Those with access can still waste time figuring out the location of the application and how to connect to it over VPN. In addition, reproducing problems in test environments takes too much time and isn't always possible. System administrators spend a lot of time helping developers track down server information. Most companies give key developers access to all of the production resources so they can help resolve application defects. The problem is only those key people have access and they become a bottleneck. They end up spending 25-50% of their time on a daily basis trying to solve application issues because they are the only ones with access. These key employees’ time is best spent on strategic new projects, not addressing application defects. This job should fall to entry level developers, provided they have access to all the information they need to troubleshoot the problems.The solution to agile application support is giving all the developers limited access to the production environment and all the server information they need to see. Some companies create their own solutions internally to collect log files, centralize errors or other things to address the problem. Some developers even have access to server monitoring or other tools. But they key is giving them access to everything they need so they can see the full picture and giving access to the whole team. Giving access to everyone scales up the application support team and creates collaboration around providing improved application support.Stackify enables agile application supportStackify has created a solution that can give all developers a secure and read only view of the entire production server environment without console or remote desktop access.They provide a web application that provides real time visibility to the important information that developers need to see. An application centric view enables them to see all of their apps across multiple datacenters and environments. They don’t need to know where the application is deployed, just the name of the application to find it and dig in to see more. All your developers can see server health, application health, log files, config files, windows event viewer, deployment history, application notes, and much more. They can receive email and text alerts when problems arise and even safely query your production databases.Stackify enables companies that do agile development to scale up their application support team by getting more team members involved. The lead developers can spend more time on new projects. Application issues can be fixed quicker than ever. Operations can spend less time helping developers collect server information. Agile application support starts with Stackify. Visit Stackify.com to learn more.

    Read the article

< Previous Page | 12 13 14 15 16 17 18 19 20 21 22 23  | Next Page >