Search Results

Search found 24132 results on 966 pages for 'non clustered index'.

Page 187/966 | < Previous Page | 183 184 185 186 187 188 189 190 191 192 193 194  | Next Page >

  • Oracle B2B Started Kit

    - by Nitesh Jain Oracle
    This post is for audience who is starting their journey with Oracle B2B 11g practice. Thought of writing this article as even i wondered as to what to do where to search while i was like you. Hope this material will help you to understand Oracle B2B closely: Oracle B2B OTN : http://www.oracle.com/technology/products/soa/b2b/index.html Download Location: http://www.oracle.com/technology/products/soa/soasuite/collateral/downloads.html#11.1.1.3.0 Installation Guide: http://download.oracle.com/docs/cd/E14571_01/doc.1111/e13925/toc.htm User Guide: http://download.oracle.com/docs/cd/E14571_01/integration.1111/e10229/toc.htm Sample Location: B2B Samples are part of SOA sample at below location: http://www.oracle.com/technology/sample_code/products/soa/index.html Developer Notes / Step by Step configuration guide: http://www.oracle.com/technology/products/soa/b2b/index.html http://www.oracle.com/technology/products/soa/b2b/collateral/B2B_TU001_EDI.pdf http://www.oracle.com/technology/products/soa/b2b/collateral/B2B_TU002_HL7.pdf http://www.oracle.com/technology/products/soa/b2b/collateral/B2B_TU003_ebxml.pdf Blog: http://blogs.oracle.com/oracleb2bgurus/ Forum: http://forums.oracle.com/forums/forum.jspa?forumID=242&start=0 This article is as per Oracle B2B 11g PS2 Release (11.1.1.3.0)

    Read the article

  • SQL SERVER – SSMS: Memory Usage By Memory Optimized Objects Report

    - by Pinal Dave
    At conferences and at speaking engagements at the local UG, there is one question that keeps on coming which I wish were never asked. The question around, “Why is SQL Server using up all the memory and not releasing even when idle?” Well, the answer can be long and with the release of SQL Server 2014, this got even more complicated. This release of SQL Server 2014 has the option of introducing In-Memory OLTP which is completely new concept and our dependency on memory has increased multifold. In reality, nothing much changes but we have memory optimized objects (Tables and Stored Procedures) additional which are residing completely in memory and improving performance. As a DBA, it is humanly impossible to get a hang of all the innovations and the new features introduced in the next version. So today’s blog is around the report added to SSMS which gives a high level view of this new feature addition. This reports is available only from SQL Server 2014 onwards because the feature was introduced in SQL Server 2014. Earlier versions of SQL Server Management Studio would not show the report in the list. If we try to launch the report on the database which is not having In-Memory File group defined, then we would see the message in report. To demonstrate, I have created new fresh database called MemoryOptimizedDB with no special file group. Here is the query used to identify whether a database has memory-optimized file group or not. SELECT TOP(1) 1 FROM sys.filegroups FG WHERE FG.[type] = 'FX' Once we add filegroup using below command, we would see different version of report. USE [master] GO ALTER DATABASE [MemoryOptimizedDB] ADD FILEGROUP [IMO_FG] CONTAINS MEMORY_OPTIMIZED_DATA GO The report is still empty because we have not defined any Memory Optimized table in the database.  Total allocated size is shown as 0 MB. Now, let’s add the folder location into the filegroup and also created few in-memory tables. We have used the nomenclature of IMO to denote “InMemory Optimized” objects. USE [master] GO ALTER DATABASE [MemoryOptimizedDB] ADD FILE ( NAME = N'MemoryOptimizedDB_IMO', FILENAME = N'E:\Program Files\Microsoft SQL Server\MSSQL12.SQL2014\MSSQL\DATA\MemoryOptimizedDB_IMO') TO FILEGROUP [IMO_FG] GO You may have to change the path based on your SQL Server configuration. Below is the script to create the table. USE MemoryOptimizedDB GO --Drop table if it already exists. IF OBJECT_ID('dbo.SQLAuthority','U') IS NOT NULL DROP TABLE dbo.SQLAuthority GO CREATE TABLE dbo.SQLAuthority ( ID INT IDENTITY NOT NULL, Name CHAR(500)  COLLATE Latin1_General_100_BIN2 NOT NULL DEFAULT 'Pinal', CONSTRAINT PK_SQLAuthority_ID PRIMARY KEY NONCLUSTERED (ID), INDEX hash_index_sample_memoryoptimizedtable_c2 HASH (Name) WITH (BUCKET_COUNT = 131072) ) WITH (MEMORY_OPTIMIZED = ON, DURABILITY = SCHEMA_AND_DATA) GO As soon as above script is executed, table and index both are created. If we run the report again, we would see something like below. Notice that table memory is zero but index is using memory. This is due to the fact that hash index needs memory to manage the buckets created. So even if table is empty, index would consume memory. More about the internals of how In-Memory indexes and tables work will be reserved for future posts. Now, use below script to populate the table with 10000 rows INSERT INTO SQLAuthority VALUES (DEFAULT) GO 10000 Here is the same report after inserting 1000 rows into our InMemory table.    There are total three sections in the whole report. Total Memory consumed by In-Memory Objects Pie chart showing memory distribution based on type of consumer – table, index and system. Details of memory usage by each table. The information about all three is taken from one single DMV, sys.dm_db_xtp_table_memory_stats This DMV contains memory usage statistics for both user and system In-Memory tables. If we query the DMV and look at data, we can easily notice that the system tables have negative object IDs.  So, to look at user table memory usage, below is the over-simplified version of query. USE MemoryOptimizedDB GO SELECT OBJECT_NAME(OBJECT_ID), * FROM sys.dm_db_xtp_table_memory_stats WHERE OBJECT_ID > 0 GO This report would help DBA to identify which in-memory object taking lot of memory which can be used as a pointer for designing solution. I am sure in future we will discuss at lengths the whole concept of In-Memory tables in detail over this blog. To read more about In-Memory OLTP, have a look at In-Memory OLTP Series at Balmukund’s Blog. Reference: Pinal Dave (http://blog.sqlauthority.com)Filed under: PostADay, SQL, SQL Authority, SQL Query, SQL Server, SQL Server Management Studio, SQL Tips and Tricks, T SQL Tagged: SQL Memory, SQL Reports

    Read the article

  • Refresh bounded taskflows across regions using Contextual Events

    - by raghu.yadav
    Usecases: 1) Data Change in left region inputText field reflect changes in right region using contextual event. example by Frank Nimphius :Value change event refresh across regions using Contextual Events 2) Select Tree node in left region reflects dependent detail form in right region using dynamic regions and Contextual Events. example by Frank Nimphius:Example6-RangeCtx.unzip More related examples: http://thepeninsulasedge.com/frank_nimphius/2008/02/07/adf-faces-rc-refreshing-a-table-ui-from-a-contextual-event/ http://www.oracle.com/technology/products/jdev/tips/fnimphius/generictreeselectionlistener/index.html http://www.oracle.com/technology/products/jdev/tips/fnimphius/syncheditformwithtree/index.html http://biemond.blogspot.com/2009/01/passing-adf-events-between-task-flow.html http://www.oracle.com/technology/products/jdev/tips/fnimphius/opentaskflowintab/index.html http://lucbors.blogspot.com/2010/03/adf-11g-contextual-event-framework.html http://thepeninsulasedge.com/blog/?cat=2 http://www.ora600.be/news/adf-contextual-events-11g-r1-ps1

    Read the article

  • Do premium domain names help us with other languages too?

    - by Fabio Milheiro
    It's commonly known that premium domains with one or two relevant keywords may help us improve our rankings in SERPS. But would it be possible that an english premium domain, for example gold.com (no, it's not mine) also helps to drive more non-english traffic (I'm talking about non-english pages ob)? Trying to make my question clear: Let's suppose that I have an english premium domain with a page like this: gold dot com/post/123/gold-is-yellow And decide to have a spanish, portuguese or french version of the site with pages like: gold dot com/es/post/123/el-oro-es-amarillo gold dot com/pt/post/123/o-ouro-e-amarelo gold dot com/fr/post/123/fsdfsdfsdf The fact that my english domain is a premium one and highly relevant for english terms, will also help me to achieve good rankings for non-english searched terms like: oro (spanish) or ouro (portuguese)?

    Read the article

  • Pure virtual or abstract, what's in a name?

    - by Steven Jeuris
    While discussing a question about virtual functions on Stack Overflow, I wondered whether there was any official naming for pure (abstract) and non-pure virtual functions. I always relied on wikipedia for my information, which states that pure and non-pure virtual functions are the general term. Unfortunately, the article doesn't back it up with a origin or references. To quote Jon Skeet's answer to my reply that pure and non-pure are the general term used: @Steven: Hmm... possibly, but I've only ever seen it in the context of C++ before. I suspect anyone talking about them is likely to have a C++ background :) Did the terms originate from C++, or were they first defined or implemented in a earlier language, and are they the 'official' scientific terms?

    Read the article

  • Solaris 11 pkg fix is my new friend

    - by user12611829
    While putting together some examples of the Solaris 11 Automated Installer (AI), I managed to really mess up my system, to the point where AI was completely unusable. This was my fault as a combination of unfortunate incidents left some remnants that were causing problems, so I tried to clean things up. Unsuccessfully. Perhaps that was a bad idea (OK, it was a terrible idea), but this is Solaris 11 and there are a few more tricks in the sysadmin toolbox. Here's what I did. # rm -rf /install/* # rm -rf /var/ai # installadm create-service -n solaris11-x86 --imagepath /install/solaris11-x86 \ -s [email protected] Warning: Service svc:/network/dns/multicast:default is not online. Installation services will not be advertised via multicast DNS. Creating service from: [email protected] DOWNLOAD PKGS FILES XFER (MB) SPEED Completed 1/1 130/130 264.4/264.4 0B/s PHASE ITEMS Installing new actions 284/284 Updating package state database Done Updating image state Done Creating fast lookup database Done Reading search index Done Updating search index 1/1 Creating i386 service: solaris11-x86 Image path: /install/solaris11-x86 So far so good. Then comes an oops..... setup-service[168]: cd: /var/ai//service/.conf-templ: [No such file or directory] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ This is where you generally say a few things to yourself, and then promise to quit deleting configuration files and directories when you don't know what you are doing. Then you recall that the new Solaris 11 packaging system has some ability to correct common mistakes (like the one I just made). Let's give it a try. # pkg fix installadm Verifying: pkg://solaris/install/installadm ERROR dir: var/ai Group: 'root (0)' should be 'sys (3)' dir: var/ai/ai-webserver Missing: directory does not exist dir: var/ai/ai-webserver/compatibility-configuration Missing: directory does not exist dir: var/ai/ai-webserver/conf.d Missing: directory does not exist dir: var/ai/image-server Group: 'root (0)' should be 'sys (3)' dir: var/ai/image-server/cgi-bin Missing: directory does not exist dir: var/ai/image-server/images Group: 'root (0)' should be 'sys (3)' dir: var/ai/image-server/logs Missing: directory does not exist dir: var/ai/profile Missing: directory does not exist dir: var/ai/service Group: 'root (0)' should be 'sys (3)' dir: var/ai/service/.conf-templ Missing: directory does not exist dir: var/ai/service/.conf-templ/AI_data Missing: directory does not exist dir: var/ai/service/.conf-templ/AI_files Missing: directory does not exist file: var/ai/ai-webserver/ai-httpd-templ.conf Missing: regular file does not exist file: var/ai/service/.conf-templ/AI.db Missing: regular file does not exist file: var/ai/image-server/cgi-bin/cgi_get_manifest.py Missing: regular file does not exist Created ZFS snapshot: 2012-12-11-21:09:53 Repairing: pkg://solaris/install/installadm Creating Plan (Evaluating mediators): | DOWNLOAD PKGS FILES XFER (MB) SPEED Completed 1/1 3/3 0.0/0.0 0B/s PHASE ITEMS Updating modified actions 16/16 Updating image state Done Creating fast lookup database Done In just a few moments, IPS found the missing files and incorrect ownerships/permissions. Instead of reinstalling the system, or falling back to an earlier Live Upgrade boot environment, I was able to create my AI services and now all is well. # installadm create-service -n solaris11-x86 --imagepath /install/solaris11-x86 \ -s [email protected] Warning: Service svc:/network/dns/multicast:default is not online. Installation services will not be advertised via multicast DNS. Creating service from: [email protected] DOWNLOAD PKGS FILES XFER (MB) SPEED Completed 1/1 130/130 264.4/264.4 0B/s PHASE ITEMS Installing new actions 284/284 Updating package state database Done Updating image state Done Creating fast lookup database Done Reading search index Done Updating search index 1/1 Creating i386 service: solaris11-x86 Image path: /install/solaris11-x86 Refreshing install services Warning: mDNS registry of service solaris11-x86 could not be verified. Creating default-i386 alias Setting the default PXE bootfile(s) in the local DHCP configuration to: bios clients (arch 00:00): default-i386/boot/grub/pxegrub Refreshing install services Warning: mDNS registry of service default-i386 could not be verified. # installadm create-service -n solaris11u1-x86 --imagepath /install/solaris11u1-x86 \ -s [email protected] Warning: Service svc:/network/dns/multicast:default is not online. Installation services will not be advertised via multicast DNS. Creating service from: [email protected] DOWNLOAD PKGS FILES XFER (MB) SPEED Completed 1/1 514/514 292.3/292.3 0B/s PHASE ITEMS Installing new actions 661/661 Updating package state database Done Updating image state Done Creating fast lookup database Done Reading search index Done Updating search index 1/1 Creating i386 service: solaris11u1-x86 Image path: /install/solaris11u1-x86 Refreshing install services Warning: mDNS registry of service solaris11u1-x86 could not be verified. # installadm list Service Name Alias Of Status Arch Image Path ------------ -------- ------ ---- ---------- default-i386 solaris11-x86 on i386 /install/solaris11-x86 solaris11-x86 - on i386 /install/solaris11-x86 solaris11u1-x86 - on i386 /install/solaris11u1-x86 This is way way better than pkgchk -f in Solaris 10. I'm really beginning to like this new IPS packaging system.

    Read the article

  • Validating sitemap with Yahoo Explorer

    - by Joel
    Hello, I have a sitemap index on my website, which I successfully validated with "Google webmasters tools". The declarations at the top are: <sitemapindex xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.sitemaps.org/schemas/sitemap/0.9 http://www.sitemaps.org/schemas/sitemap/0.9/siteindex.xsd" xmlns="http://www.sitemaps.org/schemas/sitemap/0.9"> This index lists 2 sitemaps . One of them contains "images" tags after using the http://www.google.com/schemas/sitemap-image/1.1 schema declaration. When submitting this sitemap index to Yahoo explorer, I get error: ERROR: FF_30000 “Not an accepted feed format file. Please consult the documentaiton for supported file formats.” Any ideas? Joel

    Read the article

  • PTLQueue : a scalable bounded-capacity MPMC queue

    - by Dave
    Title: Fast concurrent MPMC queue -- I've used the following concurrent queue algorithm enough that it warrants a blog entry. I'll sketch out the design of a fast and scalable multiple-producer multiple-consumer (MPSC) concurrent queue called PTLQueue. The queue has bounded capacity and is implemented via a circular array. Bounded capacity can be a useful property if there's a mismatch between producer rates and consumer rates where an unbounded queue might otherwise result in excessive memory consumption by virtue of the container nodes that -- in some queue implementations -- are used to hold values. A bounded-capacity queue can provide flow control between components. Beware, however, that bounded collections can also result in resource deadlock if abused. The put() and take() operators are partial and wait for the collection to become non-full or non-empty, respectively. Put() and take() do not allocate memory, and are not vulnerable to the ABA pathologies. The PTLQueue algorithm can be implemented equally well in C/C++ and Java. Partial operators are often more convenient than total methods. In many use cases if the preconditions aren't met, there's nothing else useful the thread can do, so it may as well wait via a partial method. An exception is in the case of work-stealing queues where a thief might scan a set of queues from which it could potentially steal. Total methods return ASAP with a success-failure indication. (It's tempting to describe a queue or API as blocking or non-blocking instead of partial or total, but non-blocking is already an overloaded concurrency term. Perhaps waiting/non-waiting or patient/impatient might be better terms). It's also trivial to construct partial operators by busy-waiting via total operators, but such constructs may be less efficient than an operator explicitly and intentionally designed to wait. A PTLQueue instance contains an array of slots, where each slot has volatile Turn and MailBox fields. The array has power-of-two length allowing mod/div operations to be replaced by masking. We assume sensible padding and alignment to reduce the impact of false sharing. (On x86 I recommend 128-byte alignment and padding because of the adjacent-sector prefetch facility). Each queue also has PutCursor and TakeCursor cursor variables, each of which should be sequestered as the sole occupant of a cache line or sector. You can opt to use 64-bit integers if concerned about wrap-around aliasing in the cursor variables. Put(null) is considered illegal, but the caller or implementation can easily check for and convert null to a distinguished non-null proxy value if null happens to be a value you'd like to pass. Take() will accordingly convert the proxy value back to null. An advantage of PTLQueue is that you can use atomic fetch-and-increment for the partial methods. We initialize each slot at index I with (Turn=I, MailBox=null). Both cursors are initially 0. All shared variables are considered "volatile" and atomics such as CAS and AtomicFetchAndIncrement are presumed to have bidirectional fence semantics. Finally T is the templated type. I've sketched out a total tryTake() method below that allows the caller to poll the queue. tryPut() has an analogous construction. Zebra stripping : alternating row colors for nice-looking code listings. See also google code "prettify" : https://code.google.com/p/google-code-prettify/ Prettify is a javascript module that yields the HTML/CSS/JS equivalent of pretty-print. -- pre:nth-child(odd) { background-color:#ff0000; } pre:nth-child(even) { background-color:#0000ff; } border-left: 11px solid #ccc; margin: 1.7em 0 1.7em 0.3em; background-color:#BFB; font-size:12px; line-height:65%; " // PTLQueue : Put(v) : // producer : partial method - waits as necessary assert v != null assert Mask = 1 && (Mask & (Mask+1)) == 0 // Document invariants // doorway step // Obtain a sequence number -- ticket // As a practical concern the ticket value is temporally unique // The ticket also identifies and selects a slot auto tkt = AtomicFetchIncrement (&PutCursor, 1) slot * s = &Slots[tkt & Mask] // waiting phase : // wait for slot's generation to match the tkt value assigned to this put() invocation. // The "generation" is implicitly encoded as the upper bits in the cursor // above those used to specify the index : tkt div (Mask+1) // The generation serves as an epoch number to identify a cohort of threads // accessing disjoint slots while s-Turn != tkt : Pause assert s-MailBox == null s-MailBox = v // deposit and pass message Take() : // consumer : partial method - waits as necessary auto tkt = AtomicFetchIncrement (&TakeCursor,1) slot * s = &Slots[tkt & Mask] // 2-stage waiting : // First wait for turn for our generation // Acquire exclusive "take" access to slot's MailBox field // Then wait for the slot to become occupied while s-Turn != tkt : Pause // Concurrency in this section of code is now reduced to just 1 producer thread // vs 1 consumer thread. // For a given queue and slot, there will be most one Take() operation running // in this section. // Consumer waits for producer to arrive and make slot non-empty // Extract message; clear mailbox; advance Turn indicator // We have an obvious happens-before relation : // Put(m) happens-before corresponding Take() that returns that same "m" for T v = s-MailBox if v != null : s-MailBox = null ST-ST barrier s-Turn = tkt + Mask + 1 // unlock slot to admit next producer and consumer return v Pause tryTake() : // total method - returns ASAP with failure indication for auto tkt = TakeCursor slot * s = &Slots[tkt & Mask] if s-Turn != tkt : return null T v = s-MailBox // presumptive return value if v == null : return null // ratify tkt and v values and commit by advancing cursor if CAS (&TakeCursor, tkt, tkt+1) != tkt : continue s-MailBox = null ST-ST barrier s-Turn = tkt + Mask + 1 return v The basic idea derives from the Partitioned Ticket Lock "PTL" (US20120240126-A1) and the MultiLane Concurrent Bag (US8689237). The latter is essentially a circular ring-buffer where the elements themselves are queues or concurrent collections. You can think of the PTLQueue as a partitioned ticket lock "PTL" augmented to pass values from lock to unlock via the slots. Alternatively, you could conceptualize of PTLQueue as a degenerate MultiLane bag where each slot or "lane" consists of a simple single-word MailBox instead of a general queue. Each lane in PTLQueue also has a private Turn field which acts like the Turn (Grant) variables found in PTL. Turn enforces strict FIFO ordering and restricts concurrency on the slot mailbox field to at most one simultaneous put() and take() operation. PTL uses a single "ticket" variable and per-slot Turn (grant) fields while MultiLane has distinct PutCursor and TakeCursor cursors and abstract per-slot sub-queues. Both PTL and MultiLane advance their cursor and ticket variables with atomic fetch-and-increment. PTLQueue borrows from both PTL and MultiLane and has distinct put and take cursors and per-slot Turn fields. Instead of a per-slot queues, PTLQueue uses a simple single-word MailBox field. PutCursor and TakeCursor act like a pair of ticket locks, conferring "put" and "take" access to a given slot. PutCursor, for instance, assigns an incoming put() request to a slot and serves as a PTL "Ticket" to acquire "put" permission to that slot's MailBox field. To better explain the operation of PTLQueue we deconstruct the operation of put() and take() as follows. Put() first increments PutCursor obtaining a new unique ticket. That ticket value also identifies a slot. Put() next waits for that slot's Turn field to match that ticket value. This is tantamount to using a PTL to acquire "put" permission on the slot's MailBox field. Finally, having obtained exclusive "put" permission on the slot, put() stores the message value into the slot's MailBox. Take() similarly advances TakeCursor, identifying a slot, and then acquires and secures "take" permission on a slot by waiting for Turn. Take() then waits for the slot's MailBox to become non-empty, extracts the message, and clears MailBox. Finally, take() advances the slot's Turn field, which releases both "put" and "take" access to the slot's MailBox. Note the asymmetry : put() acquires "put" access to the slot, but take() releases that lock. At any given time, for a given slot in a PTLQueue, at most one thread has "put" access and at most one thread has "take" access. This restricts concurrency from general MPMC to 1-vs-1. We have 2 ticket locks -- one for put() and one for take() -- each with its own "ticket" variable in the form of the corresponding cursor, but they share a single "Grant" egress variable in the form of the slot's Turn variable. Advancing the PutCursor, for instance, serves two purposes. First, we obtain a unique ticket which identifies a slot. Second, incrementing the cursor is the doorway protocol step to acquire the per-slot mutual exclusion "put" lock. The cursors and operations to increment those cursors serve double-duty : slot-selection and ticket assignment for locking the slot's MailBox field. At any given time a slot MailBox field can be in one of the following states: empty with no pending operations -- neutral state; empty with one or more waiting take() operations pending -- deficit; occupied with no pending operations; occupied with one or more waiting put() operations -- surplus; empty with a pending put() or pending put() and take() operations -- transitional; or occupied with a pending take() or pending put() and take() operations -- transitional. The partial put() and take() operators can be implemented with an atomic fetch-and-increment operation, which may confer a performance advantage over a CAS-based loop. In addition we have independent PutCursor and TakeCursor cursors. Critically, a put() operation modifies PutCursor but does not access the TakeCursor and a take() operation modifies the TakeCursor cursor but does not access the PutCursor. This acts to reduce coherence traffic relative to some other queue designs. It's worth noting that slow threads or obstruction in one slot (or "lane") does not impede or obstruct operations in other slots -- this gives us some degree of obstruction isolation. PTLQueue is not lock-free, however. The implementation above is expressed with polite busy-waiting (Pause) but it's trivial to implement per-slot parking and unparking to deschedule waiting threads. It's also easy to convert the queue to a more general deque by replacing the PutCursor and TakeCursor cursors with Left/Front and Right/Back cursors that can move either direction. Specifically, to push and pop from the "left" side of the deque we would decrement and increment the Left cursor, respectively, and to push and pop from the "right" side of the deque we would increment and decrement the Right cursor, respectively. We used a variation of PTLQueue for message passing in our recent OPODIS 2013 paper. ul { list-style:none; padding-left:0; padding:0; margin:0; margin-left:0; } ul#myTagID { padding: 0px; margin: 0px; list-style:none; margin-left:0;} -- -- There's quite a bit of related literature in this area. I'll call out a few relevant references: Wilson's NYU Courant Institute UltraComputer dissertation from 1988 is classic and the canonical starting point : Operating System Data Structures for Shared-Memory MIMD Machines with Fetch-and-Add. Regarding provenance and priority, I think PTLQueue or queues effectively equivalent to PTLQueue have been independently rediscovered a number of times. See CB-Queue and BNPBV, below, for instance. But Wilson's dissertation anticipates the basic idea and seems to predate all the others. Gottlieb et al : Basic Techniques for the Efficient Coordination of Very Large Numbers of Cooperating Sequential Processors Orozco et al : CB-Queue in Toward high-throughput algorithms on many-core architectures which appeared in TACO 2012. Meneghin et al : BNPVB family in Performance evaluation of inter-thread communication mechanisms on multicore/multithreaded architecture Dmitry Vyukov : bounded MPMC queue (highly recommended) Alex Otenko : US8607249 (highly related). John Mellor-Crummey : Concurrent queues: Practical fetch-and-phi algorithms. Technical Report 229, Department of Computer Science, University of Rochester Thomasson : FIFO Distributed Bakery Algorithm (very similar to PTLQueue). Scott and Scherer : Dual Data Structures I'll propose an optimization left as an exercise for the reader. Say we wanted to reduce memory usage by eliminating inter-slot padding. Such padding is usually "dark" memory and otherwise unused and wasted. But eliminating the padding leaves us at risk of increased false sharing. Furthermore lets say it was usually the case that the PutCursor and TakeCursor were numerically close to each other. (That's true in some use cases). We might still reduce false sharing by incrementing the cursors by some value other than 1 that is not trivially small and is coprime with the number of slots. Alternatively, we might increment the cursor by one and mask as usual, resulting in a logical index. We then use that logical index value to index into a permutation table, yielding an effective index for use in the slot array. The permutation table would be constructed so that nearby logical indices would map to more distant effective indices. (Open question: what should that permutation look like? Possibly some perversion of a Gray code or De Bruijn sequence might be suitable). As an aside, say we need to busy-wait for some condition as follows : "while C == 0 : Pause". Lets say that C is usually non-zero, so we typically don't wait. But when C happens to be 0 we'll have to spin for some period, possibly brief. We can arrange for the code to be more machine-friendly with respect to the branch predictors by transforming the loop into : "if C == 0 : for { Pause; if C != 0 : break; }". Critically, we want to restructure the loop so there's one branch that controls entry and another that controls loop exit. A concern is that your compiler or JIT might be clever enough to transform this back to "while C == 0 : Pause". You can sometimes avoid this by inserting a call to a some type of very cheap "opaque" method that the compiler can't elide or reorder. On Solaris, for instance, you could use :"if C == 0 : { gethrtime(); for { Pause; if C != 0 : break; }}". It's worth noting the obvious duality between locks and queues. If you have strict FIFO lock implementation with local spinning and succession by direct handoff such as MCS or CLH,then you can usually transform that lock into a queue. Hidden commentary and annotations - invisible : * And of course there's a well-known duality between queues and locks, but I'll leave that topic for another blog post. * Compare and contrast : PTLQ vs PTL and MultiLane * Equivalent : Turn; seq; sequence; pos; position; ticket * Put = Lock; Deposit Take = identify and reserve slot; wait; extract & clear; unlock * conceptualize : Distinct PutLock and TakeLock implemented as ticket lock or PTL Distinct arrival cursors but share per-slot "Turn" variable provides exclusive role-based access to slot's mailbox field put() acquires exclusive access to a slot for purposes of "deposit" assigns slot round-robin and then acquires deposit access rights/perms to that slot take() acquires exclusive access to slot for purposes of "withdrawal" assigns slot round-robin and then acquires withdrawal access rights/perms to that slot At any given time, only one thread can have withdrawal access to a slot at any given time, only one thread can have deposit access to a slot Permissible for T1 to have deposit access and T2 to simultaneously have withdrawal access * round-robin for the purposes of; role-based; access mode; access role mailslot; mailbox; allocate/assign/identify slot rights; permission; license; access permission; * PTL/Ticket hybrid Asymmetric usage ; owner oblivious lock-unlock pairing K-exclusion add Grant cursor pass message m from lock to unlock via Slots[] array Cursor performs 2 functions : + PTL ticket + Assigns request to slot in round-robin fashion Deconstruct protocol : explication put() : allocate slot in round-robin fashion acquire PTL for "put" access store message into slot associated with PTL index take() : Acquire PTL for "take" access // doorway step seq = fetchAdd (&Grant, 1) s = &Slots[seq & Mask] // waiting phase while s-Turn != seq : pause Extract : wait for s-mailbox to be full v = s-mailbox s-mailbox = null Release PTL for both "put" and "take" access s-Turn = seq + Mask + 1 * Slot round-robin assignment and lock "doorway" protocol leverage the same cursor and FetchAdd operation on that cursor FetchAdd (&Cursor,1) + round-robin slot assignment and dispersal + PTL/ticket lock "doorway" step waiting phase is via "Turn" field in slot * PTLQueue uses 2 cursors -- put and take. Acquire "put" access to slot via PTL-like lock Acquire "take" access to slot via PTL-like lock 2 locks : put and take -- at most one thread can access slot's mailbox Both locks use same "turn" field Like multilane : 2 cursors : put and take slot is simple 1-capacity mailbox instead of queue Borrow per-slot turn/grant from PTL Provides strict FIFO Lock slot : put-vs-put take-vs-take at most one put accesses slot at any one time at most one put accesses take at any one time reduction to 1-vs-1 instead of N-vs-M concurrency Per slot locks for put/take Release put/take by advancing turn * is instrumental in ... * P-V Semaphore vs lock vs K-exclusion * See also : FastQueues-excerpt.java dice-etc/queue-mpmc-bounded-blocking-circular-xadd/ * PTLQueue is the same as PTLQB - identical * Expedient return; ASAP; prompt; immediately * Lamport's Bakery algorithm : doorway step then waiting phase Threads arriving at doorway obtain a unique ticket number Threads enter in ticket order * In the terminology of Reed and Kanodia a ticket lock corresponds to the busy-wait implementation of a semaphore using an eventcount and a sequencer It can also be thought of as an optimization of Lamport's bakery lock was designed for fault-tolerance rather than performance Instead of spinning on the release counter, processors using a bakery lock repeatedly examine the tickets of their peers --

    Read the article

  • Collision Detection algorithms with early Collision exit

    - by Grieverheart
    I'm using collision detection in Monte Carlo simulations and at the moment I'm using GJK which is quite fast. I can't help to think it could be done even faster though. In the simulations, about 70% of the time GJK is run, it detects a collision. Thus collisions are more than non-collisions in my case. Most collision detection algorithms I know have an early non-collision exit test. Are there any collision detection algorithms that have an early collision detect instead of non-collision and could be potentially faster than GJK in case of collision?

    Read the article

  • Arrays for a heightmap tile-based map

    - by JPiolho
    I'm making a game that uses a map which have tiles, corners and borders. Here's a graphical representation: I've managed to store tiles and corners in memory but I'm having troubles to get borders structured. For the tiles, I have a [Map Width * Map Height] sized array. For corners I have [(Map Width + 1) * (Map Height + 1)] sized array. I've already made up the math needed to access corners from a tile, but I can't figure out how to store and access the borders from a single array. Tiles store the type (and other game logic variables) and via the array index I can get the X, Y. Via this tile position it is possible to get the array index of the corners (which store the Z index). The borders will store a game object and accessing corners from only border info would be also required. If someone even has a better way to store these for better memory and performance I would gladly accept that. EDIT: Using in C# and Javascript.

    Read the article

  • C++ - Constructor or Initialize Method to Startup

    - by Bob Fincheimer
    I want to determine when to do non-trivial initialization of a class. I see two times to do initialization: constructor and other method. I want to figure out when to use each. Choice 1: Constructor does initialization MyClass::MyClass(Data const& data) : m_data() { // does non-trivial initialization here } MyClass::~MyClass() { // cleans up here } Choice 2: Defer initialization to an initialize method MyClass::MyClass() : m_data() {} MyClass::Initialize(Data const& data) { // does non-trivial initialization here } MyClass::~MyClass() { // cleans up here } So to try and remove any subjectivity I want to figure out which is better in a couple of situations: Class that encapsulates a resource (window/font/some sort of handle) Class that composites resources to do something (a control/domain object) Data structure classes (tree/list/etc.) [Anything else you can think of] Things to analyze: Performance Ease of use by other developers How error-prone/opportunities for bugs [Anything else you can think of]

    Read the article

  • How to prevent access to website without SSL connection?

    - by CraigJ
    I have a website that has an SSL certificate installed, so that if I access the website using https instead of http I will be able to connect using a secure connection. However, I have noticed that I can still access the website non-securely, ie. by using http instead of https. How can I prevent people using the website in a non-secure manner? If I have a directory on the website, eg. samples/, can I prevent non-secure connections to just this directory?

    Read the article

  • CSM DX11 issues

    - by KaiserJohaan
    I got CSM to work in OpenGL, and now Im trying to do the same in directx. I'm using the same math library and all and I'm pretty much using the alghorithm straight off. I am using right-handed, column major matrices from GLM. The light is looking (-1, -1, -1). The problem I have is twofolds; For some reason, the ground floor is causing alot of (false) shadow artifacts, like the vast shadowed area you see. I confirmed this when I disabled the ground for the depth pass, but thats a hack more than anything else The shadows are inverted compared to the shadowmap. If you squint you can see the chairs shadows should be mirrored instead. This is the first cascade shadow map, in range of the alien and the chair: I can't figure out why this is. This is the depth pass: for (uint32_t cascadeIndex = 0; cascadeIndex < NUM_SHADOWMAP_CASCADES; cascadeIndex++) { mShadowmap.BindDepthView(context, cascadeIndex); CameraFrustrum cameraFrustrum = CalculateCameraFrustrum(degreesFOV, aspectRatio, nearDistArr[cascadeIndex], farDistArr[cascadeIndex], cameraViewMatrix); lightVPMatrices[cascadeIndex] = CreateDirLightVPMatrix(cameraFrustrum, lightDir); mVertexTransformPass.RenderMeshes(context, renderQueue, meshes, lightVPMatrices[cascadeIndex]); lightVPMatrices[cascadeIndex] = gBiasMatrix * lightVPMatrices[cascadeIndex]; farDistArr[cascadeIndex] = -farDistArr[cascadeIndex]; } CameraFrustrum CalculateCameraFrustrum(const float fovDegrees, const float aspectRatio, const float minDist, const float maxDist, const Mat4& cameraViewMatrix) { CameraFrustrum ret = { Vec4(1.0f, 1.0f, -1.0f, 1.0f), Vec4(1.0f, -1.0f, -1.0f, 1.0f), Vec4(-1.0f, -1.0f, -1.0f, 1.0f), Vec4(-1.0f, 1.0f, -1.0f, 1.0f), Vec4(1.0f, -1.0f, 1.0f, 1.0f), Vec4(1.0f, 1.0f, 1.0f, 1.0f), Vec4(-1.0f, 1.0f, 1.0f, 1.0f), Vec4(-1.0f, -1.0f, 1.0f, 1.0f), }; const Mat4 perspectiveMatrix = PerspectiveMatrixFov(fovDegrees, aspectRatio, minDist, maxDist); const Mat4 invMVP = glm::inverse(perspectiveMatrix * cameraViewMatrix); for (Vec4& corner : ret) { corner = invMVP * corner; corner /= corner.w; } return ret; } Mat4 CreateDirLightVPMatrix(const CameraFrustrum& cameraFrustrum, const Vec3& lightDir) { Mat4 lightViewMatrix = glm::lookAt(Vec3(0.0f), -glm::normalize(lightDir), Vec3(0.0f, -1.0f, 0.0f)); Vec4 transf = lightViewMatrix * cameraFrustrum[0]; float maxZ = transf.z, minZ = transf.z; float maxX = transf.x, minX = transf.x; float maxY = transf.y, minY = transf.y; for (uint32_t i = 1; i < 8; i++) { transf = lightViewMatrix * cameraFrustrum[i]; if (transf.z > maxZ) maxZ = transf.z; if (transf.z < minZ) minZ = transf.z; if (transf.x > maxX) maxX = transf.x; if (transf.x < minX) minX = transf.x; if (transf.y > maxY) maxY = transf.y; if (transf.y < minY) minY = transf.y; } Mat4 viewMatrix(lightViewMatrix); viewMatrix[3][0] = -(minX + maxX) * 0.5f; viewMatrix[3][1] = -(minY + maxY) * 0.5f; viewMatrix[3][2] = -(minZ + maxZ) * 0.5f; viewMatrix[0][3] = 0.0f; viewMatrix[1][3] = 0.0f; viewMatrix[2][3] = 0.0f; viewMatrix[3][3] = 1.0f; Vec3 halfExtents((maxX - minX) * 0.5, (maxY - minY) * 0.5, (maxZ - minZ) * 0.5); return OrthographicMatrix(-halfExtents.x, halfExtents.x, -halfExtents.y, halfExtents.y, halfExtents.z, -halfExtents.z) * viewMatrix; } And this is the pixel shader used for the lighting stage: #define DEPTH_BIAS 0.0005 #define NUM_CASCADES 4 cbuffer DirectionalLightConstants : register(CBUFFER_REGISTER_PIXEL) { float4x4 gSplitVPMatrices[NUM_CASCADES]; float4x4 gCameraViewMatrix; float4 gSplitDistances; float4 gLightColor; float4 gLightDirection; }; Texture2D gPositionTexture : register(TEXTURE_REGISTER_POSITION); Texture2D gDiffuseTexture : register(TEXTURE_REGISTER_DIFFUSE); Texture2D gNormalTexture : register(TEXTURE_REGISTER_NORMAL); Texture2DArray gShadowmap : register(TEXTURE_REGISTER_DEPTH); SamplerComparisonState gShadowmapSampler : register(SAMPLER_REGISTER_DEPTH); float4 ps_main(float4 position : SV_Position) : SV_Target0 { float4 worldPos = gPositionTexture[uint2(position.xy)]; float4 diffuse = gDiffuseTexture[uint2(position.xy)]; float4 normal = gNormalTexture[uint2(position.xy)]; float4 camPos = mul(gCameraViewMatrix, worldPos); uint index = 3; if (camPos.z > gSplitDistances.x) index = 0; else if (camPos.z > gSplitDistances.y) index = 1; else if (camPos.z > gSplitDistances.z) index = 2; float3 projCoords = (float3)mul(gSplitVPMatrices[index], worldPos); float viewDepth = projCoords.z - DEPTH_BIAS; projCoords.z = float(index); float visibilty = gShadowmap.SampleCmpLevelZero(gShadowmapSampler, projCoords, viewDepth); float angleNormal = clamp(dot(normal, gLightDirection), 0, 1); return visibilty * diffuse * angleNormal * gLightColor; } As you can see I am using depth bias and a bias matrix. Any hints on why this behaves so wierdly?

    Read the article

  • Can't load some domain

    - by Grzegorz
    I can't load some domain on my ubuntu 13.04: youtube.com translate.google.com drive.google.com vk.com When I ping them they are responding: PING youtube.com (217.119.79.59) 56(84) bytes of data. 64 bytes from non-registered.plix.pl (217.119.79.59): icmp_req=1 ttl=59 time=21.9 ms 64 bytes from non-registered.plix.pl (217.119.79.59): icmp_req=2 ttl=59 time=20.5 ms 64 bytes from non-registered.plix.pl (217.119.79.59): icmp_req=3 ttl=59 time=21.3 ms --- youtube.com ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 2002ms rtt min/avg/max/mdev = 20.544/21.255/21.914/0.560 ms But they don't load in any browser like firefox, chrome, lynx. They load in my media centre on XBMCbuntu.

    Read the article

  • Subterranean IL: Generics and array covariance

    - by Simon Cooper
    Arrays in .NET are curious beasts. They are the only built-in collection types in the CLR, and SZ-arrays (single dimension, zero-indexed) have their own commands and IL syntax. One of their stranger properties is they have a kind of built-in covariance long before generic variance was added in .NET 4. However, this causes a subtle but important problem with generics. First of all, we need to briefly recap on array covariance. SZ-array covariance To demonstrate, I'll tweak the classes I introduced in my previous posts: public class IncrementableClass { public int Value; public virtual void Increment(int incrementBy) { Value += incrementBy; } } public class IncrementableClassx2 : IncrementableClass { public override void Increment(int incrementBy) { base.Increment(incrementBy); base.Increment(incrementBy); } } In the CLR, SZ-arrays of reference types are implicitly convertible to arrays of the element's supertypes, all the way up to object (note that this does not apply to value types). That is, an instance of IncrementableClassx2[] can be used wherever a IncrementableClass[] or object[] is required. When an SZ-array could be used in this fashion, a run-time type check is performed when you try to insert an object into the array to make sure you're not trying to insert an instance of IncrementableClass into an IncrementableClassx2[]. This check means that the following code will compile fine but will fail at run-time: IncrementableClass[] array = new IncrementableClassx2[1]; array[0] = new IncrementableClass(); // throws ArrayTypeMismatchException These checks are enforced by the various stelem* and ldelem* il instructions in such a way as to ensure you can't insert a IncrementableClass into a IncrementableClassx2[]. For the rest of this post, however, I'm going to concentrate on the ldelema instruction. ldelema This instruction pops the array index (int32) and array reference (O) off the stack, and pushes a pointer (&) to the corresponding array element. However, unlike the ldelem instruction, the instruction's type argument must match the run-time array type exactly. This is because, once you've got a managed pointer, you can use that pointer to both load and store values in that array element using the ldind* and stind* (load/store indirect) instructions. As the same pointer can be used for both input and output to the array, the type argument to ldelema must be invariant. At the time, this was a perfectly reasonable restriction, and maintained array type-safety within managed code. However, along came generics, and with it the constrained callvirt instruction. So, what happens when we combine array covariance and constrained callvirt? .method public static void CallIncrementArrayValue() { // IncrementableClassx2[] arr = new IncrementableClassx2[1] ldc.i4.1 newarr IncrementableClassx2 // arr[0] = new IncrementableClassx2(); dup newobj instance void IncrementableClassx2::.ctor() ldc.i4.0 stelem.ref // IncrementArrayValue<IncrementableClass>(arr, 0) // here, we're treating an IncrementableClassx2[] as IncrementableClass[] dup ldc.i4.0 call void IncrementArrayValue<class IncrementableClass>(!!0[],int32) // ... ret } .method public static void IncrementArrayValue<(IncrementableClass) T>( !!T[] arr, int32 index) { // arr[index].Increment(1) ldarg.0 ldarg.1 ldelema !!T ldc.i4.1 constrained. !!T callvirt instance void IIncrementable::Increment(int32) ret } And the result: Unhandled Exception: System.ArrayTypeMismatchException: Attempted to access an element as a type incompatible with the array. at IncrementArrayValue[T](T[] arr, Int32 index) at CallIncrementArrayValue() Hmm. We're instantiating the generic method as IncrementArrayValue<IncrementableClass>, but passing in an IncrementableClassx2[], hence the ldelema instruction is failing as it's expecting an IncrementableClass[]. On features and feature conflicts What we've got here is a conflict between existing behaviour (ldelema ensuring type safety on covariant arrays) and new behaviour (managed pointers to object references used for every constrained callvirt on generic type instances). And, although this is an edge case, there is no general workaround. The generic method could be hidden behind several layers of assemblies, wrappers and interfaces that make it a requirement to use array covariance when calling the generic method. Furthermore, this will only fail at runtime, whereas compile-time safety is what generics were designed for! The solution is the readonly. prefix instruction. This modifies the ldelema instruction to ignore the exact type check for arrays of reference types, and so it lets us take the address of array elements using a covariant type to the actual run-time type of the array: .method public static void IncrementArrayValue<(IncrementableClass) T>( !!T[] arr, int32 index) { // arr[index].Increment(1) ldarg.0 ldarg.1 readonly. ldelema !!T ldc.i4.1 constrained. !!T callvirt instance void IIncrementable::Increment(int32) ret } But what about type safety? In return for ignoring the type check, the resulting controlled mutability pointer can only be used in the following situations: As the object parameter to ldfld, ldflda, stfld, call and constrained callvirt instructions As the pointer parameter to ldobj or ldind* As the source parameter to cpobj In other words, the only operations allowed are those that read from the pointer; stind* and similar that alter the pointer itself are banned. This ensures that the array element we're pointing to won't be changed to anything untoward, and so type safety within the array is maintained. This is a typical example of the maxim that whenever you add a feature to a program, you have to consider how that feature interacts with every single one of the existing features. Although an edge case, the readonly. prefix instruction ensures that generics and array covariance work together and that compile-time type safety is maintained. Tune in next time for a look at the .ctor generic type constraint, and what it means.

    Read the article

  • Create new variable or make multiple chained calls?

    - by Rodrigo
    What is the best way to get this attributes, thinking in performance and code quality? Using chained calls: name = this.product.getStock().getItems().get(index).getName(); id = this.product.getStock().getItems().get(index).getId(); Creating new variable: final item = this.product.getStock().getItems().get(index); name = item.getName(); it = item.getId(); I prefer the second way, to let the code cleaner. But I would like to see some opinions about it. Thank you!

    Read the article

  • How does the GPL static vs. dynamic linking rule apply to interpreted languages?

    - by ekolis
    In my understanding, the GPL prohibits static linking from non-GPL code to GPL code, but permits dynamic linking from non-GPL code to GPL code. So which is it when the code in question is not linked at all because the code is written in an interpreted language (e.g. Perl)? It would seem to be too easy to exploit the rule if it was considered dynamic linking, but on the other hand, it would also seem to be impossible to legally reference GPL code from non-GPL code if it was considered static! Compiled languages at least have a distinction between static and dynamic linking, but when all "linking" is just running scripts, it's impossible to tell what the intent is without an explicit license! Or is my understanding of this issue incorrect, rendering the question moot? I've also heard of a "classpath exception" which involves dynamic linking; is that not part of the GPL but instead something that can be added on to it, so dynamic linking is only allowed when the license includes this exception?

    Read the article

  • Updating Pages after migration of website

    - by DLackey
    My web site was coded in Coldfusion and over the years has obtained a good ranking. I recently migrated the front-end to a Wordpress site and wanted to know what is ideal way of updating Google and the various search engines of of the updates. For example, the home page of index.cfm is no longer valid since it's index.php. I've submitted an updated sitemap.xml file to Google. I'm sure my site will slip some while the search engines re-index my site but I'd like to try and minimize this as much as possible with the holidays coming up (my site is a service oriented site that caters to people who travel during the holidays). Right now, the old .cfm pages are still online but are re-routed to the appropriate Wordpress page (for example, about.cfm is now routed to /about/ using a cflocation tag.). Not sure if I should pull down the .cfm pages all together or leave them in place until the new pages are picked up by the search engines. Any advice would be helpful.

    Read the article

  • Files inside Alias folder not accessible

    - by John Isaacks
    In my apache2.conf I have an alias setup like this: Alias /cake/ /var/www-cake/repo <Directory /var/www-cake/repo> Order allow,deny Allow from all AllowOverride All Options +Indexes </Directory> inside the /var/www-cake/repo directory I just have 1 file that is index.php when I go to http://linux-server/cake/ I get a directory listing that shoes the index.php file. When I click on the file it takes me to http://linux-server/cake/index.php in which I get a 404 page not found error. What do I need to do to make the files accessible?

    Read the article

  • Adding included columns to indexes using SMO

    - by Greg Low
    A question came up on the SQL Down Under mailing list today about how to add an included column to an index using SMO. A quick search of the documentatio didn't seem to reveal any clues but a little investigation turned up what's needed: the IndexedColumn class has an IsIncluded property. Index i = new Index (); IndexedColumn ic = new IndexedColumn (i, "somecolumn" ); ic.IsIncluded = true ; Share this post: email it! | bookmark it! | digg it! | reddit! | kick it! | live it!...(read more)

    Read the article

  • Drawing Flowchart for function calculate a number in the Fibonacci Series

    - by truongvan
    I'm trying make Flowchart for function calculate a number in the Fibonacci Series. But It looks like not right. I don't how draw the recursive function. Please help me how to fix it. My flowchart: DIA This is my code: #include <iostream> using namespace std; long long Fibonacci(int input); int main() { cout << "Input Fibonacci Index number: "; int Index = 0; cin >> Index; cout << Fibonacci(i) << endl; return 0; } long long Fibonacci(int input) { if (input < 2) return input; else { return Fibonacci(input - 1) + Fibonacci(input - 2); } }

    Read the article

  • .htaccess language redirects with seo-friendly urls

    - by jlmmns
    How do I setup my .htaccess file to detect several languages, and redirect them to specific seo-friendly urls? Basically every url needs to go to index.php?lang=(...) So, for English language detection http://mysite.com has to go to http://mysite.com/en/ (index.php?lang=en) my .htaccess as of now (not working): RewriteEngine On RewriteCond %{HTTP:HOST} http://mysite.com/ RewriteCond %{HTTP:Accept-Language} ^en [NC] RewriteRule ^$ http://mysite.com/en/ [L,R=301] RewriteCond %{HTTP:Accept-Language} ^de [NC] RewriteRule ^$ http://mysite.com/de/ [L,R=301] RewriteCond %{HTTP:Accept-Language} ^nl [NC] RewriteRule ^$ http://mysite.com/nl/ [L,R=301] RewriteCond %{HTTP:Accept-Language} ^fr [NC] RewriteRule ^$ http://mysite.com/fr/ [L,R=301] RewriteCond %{HTTP:Accept-Language} ^es [NC] RewriteRule ^$ http://mysite.com/es/ [L,R=301] RewriteCond %{REQUEST_FILENAME} !-d RewriteCond %{REQUEST_FILENAME} !-f RewriteCond %{REQUEST_FILENAME} !-l RewriteRule ^(en|de|nl|fr|es)$ index.php?lang=$1 [L,QSA]

    Read the article

  • URL structure for content that is updated daily

    - by Brendon
    A small, simple site I am working on displays a single page with the day's best offers on it. The user is able to move back and forth between previous days. Which of the following URL structures works best? Structure 1 /index.html -- today's best offers /2013-06-29.html -- yesterday's best offers, etc. Structure 2 /index.html -- 302 redirects to /2013-06-30.html (or whatever today is) /2013-06-30.html -- today's best offers /2013-06-29.html -- yesterday's best offers, etc. I quite like structure 2 from the user's point of view (they can share content easily), but I am a bit concerned about updating the redirect from /index.html every single day -- would this perhaps have unintended SEO consequences?

    Read the article

  • Why is a linked list implementation considered linear?

    - by VeeKay
    My apologies for asking such a simple question. Instead of posting such basic question in SO, I felt that this is more apt a question here. I tried finding an answer for this but none of them are logically appealing or convincing to my understanding. Typically, computer memory is always linear. So is the term non linear used for a data structure in a logical sense? If so, to logically achieve non linearity in a linear computer memory, we use pointers. Right? In that case, if pointers are virtual implementations for achieving non linearity, Why would a data structure like linked list be considered linear if in reality the nodes are never physically adjacent?

    Read the article

  • Un-used Indexes on MDP_MATRIX Consuming Resources

    - by user702295
    Disable un-used Indexes: As much as it is recommended to create relevant indexes, it is advised not to have too many indexes on the mdp_matrix table.  Too many indexes will cause long waits on the table as indexes needs to get updated every time the table is updated.  There are many seeded indexes on mdp_matrix, every out of the box data model level has an index on the matrix table.  If a level is unused in the specific data model of the implementation, it is advisable to disable that index.  If the customer is not sure if and how indexes are utilized, the DBA can monitor all indexes.  After a few cycles of operation, the DBA should go over that list and see which indexes have not been used.  Consider disabling them. There are scripts on the net to monitor indexes or use the monitoring usage clause in the alter index statement.

    Read the article

< Previous Page | 183 184 185 186 187 188 189 190 191 192 193 194  | Next Page >