Search Results

Search found 11409 results on 457 pages for 'large teams'.

Page 265/457 | < Previous Page | 261 262 263 264 265 266 267 268 269 270 271 272  | Next Page >

  • Reformatting and version control

    - by l0b0
    Code formatting matters. Even indentation matters. And consistency is more important than minor improvements. But projects usually don't have a clear, complete, verifiable and enforced style guide from day 1, and major improvements may arrive any day. Maybe you find that SELECT id, name, address FROM persons JOIN addresses ON persons.id = addresses.person_id; could be better written as / is better written than SELECT persons.id, persons.name, addresses.address FROM persons JOIN addresses ON persons.id = addresses.person_id; while working on adding more columns to the query. Maybe this is the most complex of all four queries in your code, or a trivial query among thousands. No matter how difficult the transition, you decide it's worth it. But how do you track code changes across major formatting changes? You could just give up and say "this is the point where we start again", or you could reformat all queries in the entire repository history. If you're using a distributed version control system like Git you can revert to the first commit ever, and reformat your way from there to the current state. But it's a lot of work, and everyone else would have to pause work (or be prepared for the mother of all merges) while it's going on. Is there a better way to change history which gives the best of all results: Same style in all commits Minimal merge work ? To clarify, this is not about best practices when starting the project, but rather what should be done when a large refactoring has been deemed a Good Thing™ but you still want a traceable history? Never rewriting history is great if it's the only way to ensure that your versions always work the same, but what about the developer benefits of a clean rewrite? Especially if you have ways (tests, syntax definitions or an identical binary after compilation) to ensure that the rewritten version works exactly the same way as the original?

    Read the article

  • I am not the most logically-organized person. Do I have any chance at being a good 'low-level' programmer?

    - by user217902
    Background: I am entering college next year. I really enjoy making stuff and solving logical problems, so I'm thinking of majoring in compsci and working in software development. I hope to have the kind of job where I can work with implementing / improving algorithms and data structures on a regular basis.. as opposed to, say, a job that's purely concerned with mashing different libraries together, or 'finding the right APIs for the job'. (Hence the word 'low-level' in the title. No, I don't wish to write assembly all day.) Thing is, I've never been the most logically-sharp person. Thus far I have only worked on hobby projects, but I find that I make the silliest of errors ever so often, and it can take me ages to find it. Like anywhere between three hours to a day to locate a simple segfault, off-by-one error, or other logical mistake. (Of course, I do other things in the meantime, like browsing SO, reddit, and the like..) It's not like I'm 'new' to programming either; I first tried C++ maybe five years ago. My question is: is this normal? Should a programmer with any talent solve it in less time? Having read Spolsky's Smart and gets things done, where he talks about the large variance in programming speed, am I near the bottom of the curve, and therefore destined to work in companies that cannot afford to hire quality programmers? I'd like to think that conceptually I'm okay -- I can grasp algorithms and concepts pretty well, I do fine in math and science, although I probably drop signs in my equations more often than the next guy. Still, grokking concepts makes me happy, and is the reason why I want to work with algorithms. I'm hoping to hear from those of you with real-world programming experience. TL;DR: I make many careless mistakes, should I not consider programming as a career?

    Read the article

  • How can I create an orthographic display that handles different screen dimensions?

    - by Piku
    I'm trying to create an iPad/iPhone game using GLES2.0 that contains a 3D scene with a heads-up-display/GUI overlaid on the top. However, this problem would also apply if I were to port my game to a computer and run the game in a resizable window, or allow the user to change screen resolutions... When trying to make the 2D GUI/HUD work I've made the assumption that all I'm really doing is drawing a load of 2D textured 'quads' on the screen and am trying to treat the orthographic projection as an old-style 2D display with 0,0 in the upper left and screenWidth,ScreenHeight in the lower right. This causes me all sorts of confusion when I rotate my ipad into Landscape mode since I can't work out what to put into my projection and modelview matrices to turn everything around the right way. It also gets messy if I want to support the iPad's large screen, an iPhone or a Retina display since I have to then draw three sets of textures for everything and work out which ones to use. Should I be trying to map the 2D OpenGL co-ords 1:1 with the screen? While typing out this question it occurs to me that I could keep my origin in the centre, still running -1/+1 along the axes. This would let me scale my 2D content appropriately on the different screen sizes, but wouldn't I end up with the textures being scaled and possibly losing quality? I'm using OpenGLES 2.0 and have a matrix library that has equivalents to the GLES1.1 glOrthof() and glFrustrum() calls.

    Read the article

  • Best way: restructure an existing Team Foundation Server (TFS) solution

    - by dhh
    In my department we are developing several smaller AddOns for some unified communication server. For versioning and distributed development we use a Team Foundation Server 2012. But: there is only one large TFS solution for all of our applications and libraries: Main Solution Applications App 1 App 2 App 3 Externals Libraries Lib 1 Lib 2 Tools The "Application" path contains all main applications. Those are not depending on each other, but they depend on the Libraries and Externals projects. The "Externals" path contains some external DLLs referenced in our Applications and Libraries. The Libraries path contains commonly used libs (UI templates, Helper classes, etc.). They do not depend on each other and they are referenced in the Libraries and the Tools projects. The Tools path contains some helper programs like setup helpers, update web services, etc. Now, there's some major points why I'd like to change this structure: We can't use server builds. It's uncomfortable to manage TFS scrum management with sprints, impediments, etc. with a solution structure like that. Every developer always has access to all projects in the solution. A complete build lasts too long if one accidentally hits [F6] in Visual Studio... What would you change in this solution? How would you break those projects into smaller Solutions, how should those solutions be structured. My first approach would be, to create one TFS project for each Application, Library and Tool. But how can I ensure that e.g. App 2 always contains the newest version of Lib 1? Do I have to monitor changes on Lib 1 and update App 2 manually as soon as the Lib changes? Or can I somehow force Visual Studio to always use the newest version of an external project somehow?

    Read the article

  • Why not commit unresolved changes?

    - by Explosion Pills
    In a traditional VCS, I can understand why you would not commit unresolved files because you could break the build. However, I don't understand why you shouldn't commit unresolved files in a DVCS (some of them will actually prevent you from committing the files). Instead, I think that your repository should be locked from pushing and pulling, but not committing. Being able to commit during the merging process has several advantages (as I see it): The actual merge changes are in history. If the merge was very large, you could make periodic commits. If you made a mistake, it would be much easier to roll back (without having to redo the entire merge). The files could remain flagged as unresolved until they were marked as resolved. This would prevent pushing/pulling. You could also potentially have a set of changesets act as the merge instead of just a single one. This would allow you to still use tools such as git rerere. So why is committing with unresolved files frowned upon/prevented? Is there any reason other than tradition?

    Read the article

  • Windows Defender Update KB915597 (Definition 1.135.415.0)? Killed My Live Discs

    - by user88311
    Here's my problem, for those willing to read for about 2 minutes here's the entire story, http://answers.microsoft.com/en-us/windows/forum/windows_vista-windows_update/bsod-after-windows-defender-update-kb915597/a4b5fca3-0274-47b4-97c4-61b34c4c4599, for those who want the short version here's what happened. After windows update automatically updated windows defender to kb915597, my computer starting getting bsods on shut down and start up and started experiencing problems withe the usb ports. So I decided to go to the microsoft answers site for help, I know that was probably my first mistake, so I followed their advice and they turned my computer into a large paper weight. Luckily I make physical backups of my c drive every few months and I have one from back in july, so I figured I'd boot up a ubuntu live disc, copy all my files from the past 2 months to a external drive and just copy the backup back to the c drive, that's where I ran into this problem. When I put in either a ubuntu or kubuntu disc, everything goes well, until it finishes the loading bar then when the OS would presumably start up, the computer resets, I've tried with ubuntu, kubuntu and gparted, and only gparted is able to get to the point where it starts up, but even then when I try to access the internet from it, the computer resets, and when I tried to copy the entire C drive partition to a blank external I wasn't able to. So I figured somehow maybe the C drive had something to do with it, so I unplugged the C drive so my computer was just a 2.8 ghz processor and 2 gigs of ram, should have had no problem starting a live disc but the problem still continues. After doing some googling around I've found whenever windows gets a update with the title KB915597 it's pretty much the kill switch for windows, I've tried contacting microsoft tech support and even managed to directly contact a software engineer but as soon as I mention KB915597 they all just blew me off. I hope anybody who reads this has any idea how to fix this, I'm going to attempt to install ubuntu or kubuntu to a external drive using the same computer and see what happens now.

    Read the article

  • Can T520+Bumblebee run an external monitor via DisplayPort?

    - by Fen
    Using 64-bit Ubuntu 11.10 and integrated (Intel) graphics, I can run the 1600x900 laptop display plus a 1600x1200 external monitor connected to the VGA adapter. But my external monitor is 1920x1200 so I have black stripes on each side. I believe the resolution is limited like this as the maximum resolution available from the Intel GPU is 2560x1600 = 4,096,000 pixels and I'm asking for a 3520x1200 = 4,224,000 display (with 1200x300 lost above the laptop screen). At 3200x1200 = 3,840,000 pixels, the Intel GPU seems happy. Under Windows, the same limit exists when using the VGA adapter, but if I turn Optimus on then I can connect the external monitor to the DisplayPort and get its full resolution and an extended desktop. I've seen that Bumblebee can run apps on the DisplayPort using the 'optirun' command. My question is: can Bumblebee run the DisplayPort in concert with the Intel card running the laptop screen creating a large virtual desktop (as on Windows)? If so, are there any pointers to how to do this? I tried once, failed, and dropped back to Integrated Graphics (and black stripes) as I could find no reports of this configuration working.

    Read the article

  • Conventions for search result scoring

    - by DeaconDesperado
    I assume this type of question is more on-topic here than on regular SO. I have been working on a search feature for my team's web application and have had a lot of success building a multithreaded, "divide and conquer" processing system to work through a large amount of fulltext. Our problem domain is pretty specific. Users of the app generate posts, and as a general rule, posts that are more recent are considered to be of greater relevance. Some of the data we are trying to extract from search is very specific (user's feelings about specific items or things) and we are using python nltk to do named-entity extraction to find interesting likely query terms. Essentially we look for descriptive adjective-noun pairs and generate a general picture of a user's expressed sentiment as a list of tokens. This search is intended as an internal tool for our team to draw out a local picture of sentiments like "soggy pizza." There's some machine learning in there too to do entity resolution on terms like "soggy" to all manner of adjectives expressing nastiness. My problem is I am at a loss for how to go about scoring these results. The text being searched is split up into tokens in a list, so my initial approach would be to normalize a float score between 0.0-1.0 generated off of how far into the list the terms appear and how often they are repeated (a later mention of the term being worth less, earlier more, greater frequency-greater score, etc.) A certain amount of weight could be given to the timestamp as well, though I am not certain how to calculate this. I am curious if anyone has had to solve a similar problem in a search relevance grading between appreciable metrics (frequency, term location/colocation, recency) and if there are and guidelines for how to weight each. I should mention as well that the final fallback procedure in the search is to pipe the query to Sphinx, which has its own scoring practices. Sphinx operates as the last resort in case our application specific processing can't find any eligible candidates.

    Read the article

  • Unit testing to prove balanced tree

    - by Darrel Hoffman
    I've just built a self-balancing tree (red-black) in Java (language should be irrelevant for this question though), and I'm trying to come up with a good means of testing that it's properly balanced. I've tested all the basic tree operations, but I can't think of a way to test that it is indeed well and truly balanced. I've tried inserting a large dictionary of words, both pre-sorted and un-sorted. With a balanced tree, those should take roughly the same amount of time, but an unbalanced tree would take significantly longer on the already-sorted list. But I don't know how to go about testing for that in any reasonable, reproducible way. (I've tried doing millisecond tests on these, but there's no noticeable difference - probably because my source data is too small.) Is there a better way to be sure that the tree is really balanced? Say, by looking at the tree after it's created and seeing how deep it goes? (That is, without modifying the tree itself by adding a depth field to each node, which is just wasteful if you don't need it for anything other than testing.)

    Read the article

  • Dim (NEARLY blank) laptop screen, secondary screen works - why?

    - by LIttle Ancient Forest Kami
    My laptop screen is (almost) black while my secondary screen is fine. I believe it to be backlight / brightness related. Problem description it starts when I start the laptop system loads and works fine, just screen has problems I can see the screen though very faintly / dimly - it's hard to see anything which ain't very white e.g. starting screen has big Thinkpad logo in white, large font - I can see it, though very dimly second screen works very well Official backligtht debugging: using acpi setting as prescribed there for Thinkpads didn't help I can see an entry in /sys/class/backlight/ and it changes when I press hotkeys for brightness (current backlight power for instance goes up or down) acpi-off didn't helpm neither did acpi_backlight=vendor Hardware data Laptop is Thinkpad Edge with glossy screen. 4 processors, 2 cores, exemplary CPU data from cat /proc/cpuinfo reports Genuine Intel i5 (M 480 @ 2.67GHz). OS is Ubuntu Lucid, 10.04 LTS, 64-bit, with Linux generic kernel (2.6.32-44) and GNOME 2.32.2 (though I doubt there lies the problem). $ lspci | grep VGA 01:00.0 VGA compatible controller: ATI Technologies Inc M92 [Mobility Radeon HD 4500 Series] $ lshw -C display *-display description: VGA compatible controller product: M92 [Mobility Radeon HD 4500 Series] vendor: ATI Technologies Inc physical id: 0 bus info: pci@0000:01:00.0 version: 00 width: 32 bits clock: 33MHz capabilities: pm pciexpress msi bus_master cap_list rom configuration: driver=radeon latency=0 resources: irq:33 memory:c0000000-dfffffff(prefetchable) ioport:2000(size=256) memory:f0300000-f030ffff memory:f0320000-f033ffff(prefetchable) Driver I was NOT running any proprietary drivers, just checked with "Hardware drivers". There is one for ATI that is suggested there, though I didn't need it so far. UPDATE: changing the driver to proprietary one (ATI/AMD FGLRX) didn't help. Tried and failed Resetting / running on power or battery / charging / getting rid of static electricity / warming up *doesn't help* This is NOT a blank-screen problem, at least it isn't following official Ubuntu black-screen diagnostics - I can see my screen, though barely. What I will try next: - check last updates I've made - IIRC I am running on nomodeset already, but will verify this Any ideas how to proceed best? What is most probable cause?

    Read the article

  • Over-scan Issues when using HDTV through VGA

    - by RPG Master
    Right now all we can do is set the TV to 1280x768 instead of its native resolution of 1360x768. Setting it to its native resolution gives you a screen with a large portion of the left side of the screen cut off. We've tried everything with the TV so now we're turning to the innards of Ubuntu in hopes of fixing this. The computer is using an NVIDIA GeForce GT240. This is its current xorg.conf: # nvidia-settings: X configuration file generated by nvidia-settings # nvidia-settings: version 1.0 (buildd@palmer) Fri Apr 9 10:35:18 UTC 2010 Section "ServerLayout" Identifier "Layout0" Screen 0 "Screen0" 0 0 InputDevice "Keyboard0" "CoreKeyboard" InputDevice "Mouse0" "CorePointer" Option "Xinerama" "0" EndSection Section "Files" EndSection Section "InputDevice" # generated from default Identifier "Mouse0" Driver "mouse" Option "Protocol" "auto" Option "Device" "/dev/psaux" Option "Emulate3Buttons" "no" Option "ZAxisMapping" "4 5" EndSection Section "InputDevice" # generated from default Identifier "Keyboard0" Driver "kbd" EndSection Section "Monitor" # HorizSync source: builtin, VertRefresh source: builtin # HorizSync 28.0 - 55.0 # VertRefresh 43.0 - 72.0 Identifier "Monitor0" VendorName "Unknown" ModelName "CRT-0" HorizSync 28.0 - 55.0 VertRefresh 43.0 - 72.0 Option "DPMS" EndSection Section "Device" Identifier "Device0" Driver "nvidia" VendorName "NVIDIA Corporation" BoardName "GeForce 6600" EndSection Section "Screen" # Removed Option "metamodes" "1360x768 +0+0; 800x600 +0+0" Identifier "Screen0" Device "Device0" Monitor "Monitor0" DefaultDepth 24 Option "TwinView" "0" Option "TwinViewXineramaInfoOrder" "CRT-0" Option "metamodes" "1360x768 +0+0" SubSection "Display" Depth 24 EndSubSection EndSection

    Read the article

  • Is there a common programming term for the problems of adding features to an already-featureful program?

    - by Jeremy Friesner
    I'm looking for a commonly used programming term to describe a software-engineering phenomenon, which (for lack of a better way to describe it) I'll illustrate first with a couple of examples-by-analogy: Scenario 1: We want to build/extend a subway system on the outskirts of a small town in Wyoming. There are the usual subway-problems to solve, of course (hiring the right construction company, choosing the best route, buying the subway cars), but other than that it's pretty straightforward to implement the system because there aren't a huge number of constraints to satisfy. Scenario 2: Same as above, except now we need to build/extend the subway system in downtown Los Angeles. Here we face all of the problems we did in case (1), but also additional problems -- most of the applicable space is already in use, and has a vocal constituency which will protest loudly if we inconvenience them by repurposing, redesigning, or otherwise modifying the infrastructure that they rely on. Because of this, extensions to the system happen either very slowly and expensively, or they don't happen at all. I sometimes see a similar pattern with software development -- adding a new feature to a small/simple program is straightforward, but as the program grows, adding further new features becomes more and more difficult, if only because it is difficult to integrate the new feature without adversely affecting any of the large number of existing use-cases or user-constituencies. (even with a robust, adaptable program design, you run into the problem of the user interface becoming so elaborate that the program becomes difficult to learn or use) Is there a term for this phenomenon?

    Read the article

  • Is there a schematic overview of Ubuntu's architecture?

    - by joebuntu
    Hi there, as enthusiastic, advanced Linux learner, I'd love to get an overview about Linux' architecure/structure in general. You know, like "the big picture". I'm thinking of a large schematic graphic showing what is what, who is who, what system (e.g. X) comprises which subsystems (GDM/Gnome/Compiz) on the way from a to z, from boot to interactive desktop, including the most important background services (auth, network, cron, ...). Maybe a bit like this: http://www.flickr.com/photos/pgc/140859386/ but way more detailed. There's bootchart, which produces very comprehensive charts, but they again are too detailed and difficult to get the "big picture" from. Is there such a thing? Possibly not for the whole System, but maybe for single subsystems? I had trouble searching for this, because using search terms like "scheme" or "architecture" pointed to the wrong direction (a tool called "scheme" or CAD software for linux). I appreciate any links. If there's interest in those schematic overviews and links, maybe someone could turn this post into a wiki post? Cheers, joebuntu

    Read the article

  • Dynamic quadrees

    - by paul424
    recently I come out writing Quadtree for creatures culling in Opendungeons game. Thing is those are moving points and bounding hierarchy will quickly get lost if the quadtree is not rebuild very often. I have several variants, first is to upgrade the leaf position , every time creature move is requested. ( note if I would need collision detection anyway, so this might be necessery anyway). Second would be making leafs enough large , that the creature would sure stay inside it's bounding box ( due to its speed limit). The partition of a plane in quadtree is always fixed ( modulo the hierarchical unions of some parts) . For creatures close to the center of the plane , there would be no way of keeping it but inside one big leaf, besides this brokes the invariant that each point can be put into any small area as desired. So on the second thought could I use several quadrees ? Each would have its "coordinate axis XY" somwhere shifted ? Before I start playing with this maybe some other space diving structure would suit me better, unfortunetly wiki does not compare it's execution time : http://en.wikipedia.org/wiki/Grid_%28spatial_index%29#See_also

    Read the article

  • Install Windows7 on drive with Ubuntu 12.04 already on. Is my plan good?

    - by John F
    I have Ubuntu 12.04 working fine, but need W7 occasionally. I just wanted to check that my plan for installing would work? Any help appreciated. Current partitions are: Partition....@ File System @ Mount Point @ Size.....@ Used.....@ Flags /dev/sda1....@ ext4........@ /ext4a......@ 37 GiB...@ 776 MiB..@ boot /dev/sda2....@ extended....@.............@ 122 GiB..@ -........@ ./dev/sda5...@ ext4........@ / ..........@ 37 GiB...@ 6 GiB....@ .unallocated @ unallocated @.............@ 7 GiB....@ - ...... @ ./dev/sda6 ..@ ext4........@ /home.......@ 77 GiB...@ 32 GiB...@ .unallocated @ unallocated @.............@ 65 GiB...@ - .......@ /dev/sda3...@ linux-swap..@.............@ 7 GiB....@ - .......@ My plan is to: - boot to ubuntu from USB ISO - change sda1 to NTFS - install W7 to sda1 - use the "Master Boot Record repair" utility to configure dual boot so I can see my original ubuntu installation as well as W7. Have I missed something? I'm concerned as to what the 776MB is that will be overwritten by the change to NTFS. It seems large for just the MBR? Would also appreciate it if anyone can explain what sda5 and 6 are being used for? Is sda5 Ubuntu and sda6 my data? Thanks in advance.

    Read the article

  • Android - big game universe

    - by user1641923
    I am new to an Android development, though I have much experience with Java, C++, PHP programming and a bit experience with vector graphics too (basic 3d Studio Max, Flash, etc). I am starting to work on an Android game. It is going to be a 2D space shooter/RPG, and I am not going to use any game engines and any 3D party libs. I really want to create a very large game universe, or even pseudo-infinite (without visible borders, as if it were a 2D projection of a sphere). It should include 10-12 clusters of 7-8 planets/other space objects and random amount of single asteroids/comets, which player can interact with and also not interactive background. I am looking for a least complicated aproach to create such a universe. My current ideas are: Simply create bitmaps with space scenery background so that they can be tiled seamlessly repeated and construct my 2D universe of this tiles, then place interactive objects (planets, other spaceships) on it. Using vector graphics. I would have a solid color background, some random background objects and gradients here and there. My problems here: Lack of knowledge of how well vector graphics is integrated in Android. Performance? Memory usage? Does Android manage big bitmaps well? Do all of the bitmaps have to be in memory during all game process? I am interested in technical details regarding each of the ideas and a suggestion, which I should go with.

    Read the article

  • Is it appropriate to run a complex enterprise-system configuration and migration project in a similar way to a Scrum development project?

    - by AndyM
    I'm just starting out on the implementation of a large enterprise-wide system, which has complex requirements and many stakeholders. The company has been through high-level evaluation and tender process and determined to purchase a highly configurable "off-the-shelf" product rather than building an entirely bespoke system. The system will replace several existing systems and will require a significant amount of data migration. I'm thinking that the implementation of this system (which is expected to take over 2 years) could be run in a similar way to a Scrum software development project. With the first sprints targeted at building the minimal possible functionality needed (across all functional areas), and then iteratively deepening the level of functionality according the stakeholder feedback. I think this will de-risk the project and help ensure a balance of stakeholder needs within the available time. The user stories are still the same, it's just that to implement them we have work within the constraints of the pre-purchased system. When it comes to 'building stuff', instead of writing custom code the team will be configuring the off-the-shelf package, writing data conversion scripts and the like (and it should be a lot quicker!). Does this sound like a sensible approach? Does the Agile approach makes sense here?

    Read the article

  • How to Create a Grid for a 2D Game?

    - by SoulBeaver
    So I'm currently writing the engine for my videogame. I've almost integrated Tiled (I think) so I should have a map-creator here soon. My question is, how do I actually make the grid? I'm really confused here. If I create a large map with, say, 20x20 grids the size of 32x32 (screen size 640x640), then what do I do with it? Let's say I have the code for creating a window, and then place a player sprite that I can move with input, that's fine. If I use one map that's as big as the screen, then every pixel on the map is also a pixel on the game screen. The mapping is exact. Now what happens if I have a 2000x2000 map, for example? My character would have to keep moving and move the map around (or rather the camera focused on the player moves). Then I can no longer say that the screen maps exactly to the pixel position of the map. I tried making a Grid class that maps out the screen area to 32x32 tiles, but I'm not sure if that makes any sense. Once the map moves each tile would have to update its information, or something. I'm just really confused here. How do I actually make the tiles and a grid and map them to the data I get from tiled, or that I make myself? Are there any good examples of source code that I could look at?

    Read the article

  • Turn Based Event Algorithm

    - by GamersIncoming
    I'm currently working on a small roguelike in XNA, which sees the player in a randomly generated series of dungeons fending off creeps, as you might expect. As with most roguelikes, the player makes a move, and then each of the creeps currently active on screen will make a move in turns, until all creeps have updated, and it return's to the player's go. On paper, the simple algorithm is straightforward: Player takes turn Turn Number increments For each active creep, update Position Once all active creeps have updated, allow player to take next turn However, when it comes to actually writing this in more detail, the concept becomes a bit more tricky for me. So my question comes as this: what is the best way to handle events taking turns to trigger, where the completion of each last event triggers the next, when dealing with a large number of creeps (probably stored as an array of an enemy object), and is there an easier way to create some kind of engine that just takes all objects that need updating and chains them together so their updates follow suit? I'm not asking for code, just algorithms and theory in the direction of objects triggering updates one after the other, in a turn based manner. Thanks in advance. Edited: Here's the code I currently have that is horrible :/ if (player.getTurnOver() && updateWait == 0) { if (creep[creepToUpdate].getActive()) { creep[creepToUpdate].moveObject(player, map1); updateWait = 10; } if (creepToUpdate < creep.Length -1) { creepToUpdate++; } else { creepToUpdate = 0; player.setTurnOver(false); } } if (updateWait > 0) { updateWait--; }

    Read the article

  • Basic Puppet installation with Solaris 11.2 beta

    - by user13366125
    At the recent announcement we talked a lot about the Puppet integration. But how do you set it up? I want to show this in this blog entry. However this example i'm using is even useful in practice. Due to the extremely low overhead of zones i'm frequently seeing really large numbers of zones on a single system. Changing /etc/hosts or changing an SMF service property on 3 systems is not that hard. Doing it on a system with 500 zones is ... let say it diplomatic ... a job you give to someone you want to punish. Puppet can help in this case making of managing the configuration and to ease the distribution. You describe the changes you want to make in a file or set of file called manifest in the Puppet world and then roll them out to your servers, no matter if they are virtual or physical. A warning at first: Puppet is a really,really vast topic. This article is really basic and it doesn't goes more than just even toe's deep into the possibilities and capabilities of Puppet. It doesn't try to explain Puppet ... just how you get it up and running and do basic tests. There are many good books on Puppet. Please read one of them, and the concepts and the example will get much clearer immediately. (more)

    Read the article

  • When NOT to use a framework

    - by Chris
    Today, one can find a framework for just about any language, to suit just about any project. Most modern frameworks are fairly robust (generally speaking), with hour upon hour of testing, peer reviewed code, and great extensibility. However, I think there is a downside to ANY framework in that programmers, as a community, may become so reliant upon their chosen frameworks that they no longer understand the underlying workings, or in the case of newer programmers, never learn the underlying workings to begin with. It is easy to become specialized to a degree that you are no longer a 'PHP programmer' (for example), but a "Drupal programmer", to the exclusion of anything else. Who cares, right? We have the framework! We don't need to know how to "do it by hand"! Right? The result of this loss of basic skills (sometimes to the extent that programmers who don't use frameworks are viewed as "outdated") is that it becomes common practice to use a framework where it is not required or appropriate. The features the framework facilitates wind up confused with what the base language is capable of. Developers start using frameworks to accomplish even the most basic of tasks, so that what once was considered a rudimentary process now involves large libraries with their own quirks, bugs, and dependencies. What was once accomplished in 20 lines is now accomplished by including a 20,000 line framework AND writing 20 lines to use the framework. Conversely, one does not want to reinvent the wheel. If I'm writing code to accomplish some basic, common little task, I might feel like I am wasting my time when I know that framework XYZ offers all the features I am after, and a whole lot more. The "whole lot more" part still has me worried, but it doesn't seem that many even consider it anymore. There has to be a good metric to determine when it is appropriate to use a framework. What do you consider the threshold to be, how do you decide when to use a framework, or, when not.

    Read the article

  • I just received a complaint from a user of the website I maintain. Should I do anything?

    - by Chris
    I was sent sent a large wall of text from a user of the website I maintain at my job. They are clearly upset for having to deal with a horribly outdated web application that has not seen any serious updates in over 6+ years. No refactoring has been done, the code quality is terrible, the security unchecked, policy compliances ignored, in addition to being ugly and frankly embarrassing. Keep in mind this is a small business but the website is used by hundreds daily. I'm one of two programmers there, and I've been working there for two years. This person says they are about my age (22) and understand technology (but can't use proper grammar). The complaint mentioned awkward pages and actions on the website, but they don't even have a clue as to the depth of the flaws in this website. Now, I would love to honestly tell them that there's a lot wrong with this company and that this application was built when we were in high school. And that while it's not my fault that the website is terrible, I'm the one in position to fix it. But on the other hand, I could just say nothing and ignore it. Would doing this publicly have any advantage to future employees (showing integrity) or would it just be a completely pointless mistake? Odds are, even if I respond only that one person will ever read it. Regardless, I'm probably just going to ignore it and continue starting my project to refactor the website.

    Read the article

  • Using a random string to authenticate HMAC?

    - by mrwooster
    I am designing a simple webservice and want to use HMAC for authentication to the service. For the purpose of this question we have: a web service at example.com a secret key shared between a user and the server [K] a consumer ID which is known to the user and the server (but is not necessarily secret) [D] a message which we wish to send to the server [M] The standard HMAC implementation would involve using the secret key [K] and the message [M] to create the hash [H], but I am running into issues with this. The message [M] can be quite long and tends to be read from a file. I have found its very difficult to produce a correct hash consistently across multiple operating systems and programming languages because of hidden characters which make it into various file formats. This is of course bad implementation on the client side (100%), but I would like this webservice to be easily accessible and not have trouble with different file formats. I was thinking of an alternative, which would allow the use a short (5-10 char) random string [R] rather than the message for autentication, e.g. H = HMAC(K,R) The user then passes the random string to the server and the server checks the HMAC server side (using random string + shared secret). As far as I can see, this produces the following issues: There is no message integrity - this is ok message integrity is not important for this service A user could re-use the hash with a different message - I can see 2 ways around this Combine the random string with a timestamp so the hash is only valid for a set period of time Only allow each random string to be used once Since the client is in control of the random string, it is easier to look for collisions I should point out that the principle reason for authentication is to implement rate limiting on the API service. There is zero need for message integrity, and its not a big deal if someone can forge a single request (but it is if they can forge a very large number very quickly). I know that the correct answer is to make sure the message [M] is the same on all platforms/languages before hashing it. But, taking that out of the equation, is the above proposal an acceptable 2nd best?

    Read the article

  • One good reason for a rewrite

    - by Supermighty
    I have a personal web project I cut my teeth on learning how to program. I wrote it in PHP and learned as I went. I eventually I re-factored it to use MVC and removed all mixing of php/html. Right now it has no users, save myself, and it makes no money. I have a strong desire to rewrite the entire app. Which really isn't that large of an app. I have a lot of reasons why I should not rewrite it. I know that I should move forward. It's a working app now and it will only set me back to rewrite it. But I can't shake this feeling that I would be better off using a different programming language in the long run. That I'd enjoy it more. That I'd feel comfortable with it. I feel like my one good reason to rewrite my app is that I have a gut feeling that I should. PHP seems like a hack thrown together. I want to use a language that feels more elegant to me. Any feedback you have would be welcome.

    Read the article

  • Updated 12.04 and now Xorg is dead from NVidia drivers

    - by Xatolos
    I updated (August 16) Ubuntu 12.04 today and while I admit I didn't look what was being updated, it was large enough to have had an NVidia update in it. Problem is later on when I turned on my laptop again it won't boot into a GUI, just to the terminal (which I don't really know much of how to do anything). Now I have been able to log in and type "startx" to start up the GUI and now its complaining that my NVidia driver (if I remember correctly) is listed as being version 304.(number) and the version installed/found is version 295.(number). I don't know how this happened and now I'm stuck. I have no idea how it happened and don't know how to fix. Does anyone have any ideas? I have only access to wifi internet and no idea how to log on without the GUI so getting a new NVidia package isn't something I would know how to do even... Any and all help would be highly appreciated. edit: If it helps, I'm using a ASUS G Series G53JW-A1 http://www.newegg.com/Product/Product.aspx?Item=N82E16834220861

    Read the article

< Previous Page | 261 262 263 264 265 266 267 268 269 270 271 272  | Next Page >