Search Results

Search found 14602 results on 585 pages for 'objected oriented design'.

Page 109/585 | < Previous Page | 105 106 107 108 109 110 111 112 113 114 115 116  | Next Page >

  • Learning to write organized and modular programs

    - by Peter
    I'm a computer science student, and I'm just starting to write relatively larger programs for my coursework (between 750 - 1500 lines). Up until now, it's been possible to get by with any reasonable level of modularization and object oriented design. However, now that I'm writing more complex code for my assignments I'd like to learn to write better code. Can anyone point me in the direction of some resources for learning about what sort of things to look for when designing your program's architecture so that you can make it as modularized as possible?

    Read the article

  • Does Scheme work with Microsoft COM?

    - by Martin
    I'm new to Scheme -- the functional programming language and I like it a lot for its first-class/higher-order functions. However, my data comes from a COM source with an object-oriented API. I know Scheme and COM belong to different programming paradigms, but I'm wondering if there is any interface or a way for Scheme to connect to a COM source? Thanks.

    Read the article

  • Objects With No Behavior

    - by Patrick Donovan
    I've been teaching myself object oriented programming and I'm thinking about a situation where I have an object "Transaction", that has quite a few properties to it like account, amount, date, currency, type, etc. I never plan to mutate these data points, and calculation logic will live in other classes. My question is, is it poor Python design to instantiate thousands of objects just to hold data? I find the data far easier to work with embedded in a class rather than trying to cram it into some combination of data structures.

    Read the article

  • What question(s) does an object's behavior answer?

    - by Corwin
    Reading a book I have found the following statement: (Object) Behaviors answer either of two questions: What does this object do (for me)? or What can I do to this object? In the case of an orange, it doesn’t do a whole lot, but we can do things to it. One behavior is that it can be eaten. In my understanding of object behaviour the statement above is correct regarding the first question and is incorrect in case of the second. However, I often see classes with methods like Orange::eat(), and this makes me uncertain about my design skills. So I would like to ask is it a design mistake to give oranges a behaviour eat? (oranges and eat are used just for example)

    Read the article

  • a completely decoupled OO system ?

    - by shrini1000
    To make an OO system as decoupled as possible, I'm thinking of the following approach: 1) we run an RMI/directory like service where objects can register and discover each other. They talk to this service through an interface 2) we run a messaging service to which objects can publish messages, and register subscription callbacks. Again, this happens through interfaces 3) when object A wants to invoke a method on object B, it discovers the target object's unique identity through #1 above, and publishes a message on the message service for object B 4) message services invokes B's callback to give it the message 5) B processes the request and sends the response for A on message service 6) A's callback is called and it gets the response. I feel this system is as decoupled as practically possible, but it has the following problems: 1) communication is typically asynchronous 2) hence it's non real time 3) the system as a whole is less efficient. Are there any other practical problems where this design obviously won't be applicable ? What are your thoughts on this design in general ?

    Read the article

  • Best way for an external (remote) graphics designer to style ASP.NET MVC 4 app?

    - by Tom K
    My customer has his own graphics designer he wants to use to style his web application we're building in ASP.NET MVC 4. Our solution is in Bitbucket, but if he can't run it what choices do we have? I doubt he uses Visual Studio 2012. One idea is for us to publish to our solution to a file system, send it to him, have him create a local IIS website on his machine (assuming he isn't using a Mac). Mocking data or pointing to a test SQL in Azure isn't a problem. Then he can make changes to .css and .cshtml files. Will this even work? The point is that he needs to be able to test his changes. I know he can modify the views and just check-in. But he needs to deliver a working design. So it seems inefficient. The graphics designer will have access to our test site so he can see how it works, what data we have and fields. Another idea is for him to build a static mock site using just HTML/CSS. Later I'd integrate his styles into customer's solution, split his html into partial views which we use and add Razor syntax. Again, we'd like to leverage graphics designer for all of this. Is there a best practice documented around this subject? How do other teams deal with this situation?

    Read the article

  • Question about API and Web application code sharing

    - by opendd
    This is a design question. I have a multi part application with several user types. There is a user client for the patient that interacts with a web service. There is an API evolving behind the web service that will be exposed to institutional "users" and an interface for clinicians, researchers and admin types. The patient UI is Flex. The clinician/admin portion of the application is RoR. The API is RoR/rack based. The web service component is Java WS. All components access the same data source. These components are deployed as separate components to their own subdomains. This decision was made to allow for scaling the components individually as needed. Initially, the decision was made to split the code for the RoR Web application from the RoR API. This decision was made in the interests of security and keeping the components focused on specific tasks. Over the course of time, there is necessarily going to be overlap and I am second guessing my decision to keep the code totally separate. I am noticing code being lifted from the admin side being lifted, modified and used in the API. This being the case, I have been considering merging the Ruby based repositories. I am interested in ideas and insight on this situation along with the reasoning behind your thoughts. Thanks.

    Read the article

  • Do you need all that data?

    - by BuckWoody
    I read an amazing post over on ars technica (link: http://arstechnica.com/science/news/2010/03/the-software-brains-behind-the-particle-colliders.ars?utm_source=rss&utm_medium=rss&utm_campaign=rss) abvout the LHC, or as they are also known, the "particle colliders". Beyond just the pure scientific geek awesomeness, these instruments have the potential to collect more data than you can (or possibly should) store. Actually, this problem has a lot in common with a BI system. There's so much granular detail available in the source systems that a designer has to decide how, and how much, to roll up the data. Whenver you do that, you lose fidelity, but in many cases that's OK. Take, for example, your car's speedometer. You don't actually need to track each and every point of speed as it happens. You only need to know that you're hovering around the speed limit at a certain point in time. Since this is the way that humans percieve data, is there some lesson we should take in the design of data "flows" - and what implications does this have for new technologies like StreamInsight? Share this post: email it! | bookmark it! | digg it! | reddit! | kick it! | live it!

    Read the article

  • Game Components, Game Managers and Object Properties

    - by George Duckett
    I'm trying to get my head around component based entity design. My first step was to create various components that could be added to an object. For every component type i had a manager, which would call every component's update function, passing in things like keyboard state etc. as required. The next thing i did was remove the object, and just have each component with an Id. So an object is defined by components having the same Ids. Now, i'm thinking that i don't need a manager for all my components, for example i have a SizeComponent, which just has a Size property). As a result the SizeComponent doesn't have an update method, and the manager's update method does nothing. My first thought was to have an ObjectProperty class which components could query, instead of having them as properties of components. So an object would have a number of ObjectProperty and ObjectComponent. Components would have update logic that queries the object for properties. The manager would manage calling the component's update method. This seems like over-engineering to me, but i don't think i can get rid of the components, because i need a way for the managers to know what objects need what component logic to run (otherwise i'd just remove the component completely and push its update logic into the manager). Is this (having ObjectProperty, ObjectComponent and ComponentManager classes) over-engineering? What would be a good alternative?

    Read the article

  • Cost to licence characters or ships for a game

    - by Michael Jasper
    I am producing a game pitch document for a university game design class, and I am looking for examples of licencing cost for using characters or ships from other IP holders in a game. For example: cost of using an X-Wing in a game, licencing from Lucas cost of using the Enterprise in a game, licencing from Paramount cost of using the Space Shuttle (if any), licencing from Nasa EDIT The closest information I can find is from an article about Nights of the Old Republic, but isn't nearly specific enough for my needs: What Kotick means by Lucas being the principal beneficiary of the success of The Old Republic is that there are most likely clauses in the license agreement that give percentages, points, or another denomination of revenue out to Lucas and his people just for the Star Wars name, and that amount is presumed to be a great deal of money. Kotick is saying that because the cost of the license is so prohibitive, as he has personally had experience with in his position as CEO of Activision Blizzard, that EA will not be able to be profitable because of the hemorrhaging of money to the licensor. EDIT 2 Another vague source stating that FOX uses a "five-figure rule" (assuming between $10,000 - $99,000) It seems FOX, like most studios, will not license individuals to create new works based upon their products. They will only commission individuals of their choosing if they elect to branch out into expanded product lines related to those licenses. Alternately, they are open to making the licencing available to large corporations with access to global markets, but only if those corporations agree to what Ms Friedman called a "five-figure guarantee". Presumably this means that the corporation seeking the licensing must agree to pay a 5-figure sum for that license, and be confident that their product will sell enough volume to recoup that fee, and to produce sufficient profits to make the acquisition worth their while. Thank you!

    Read the article

  • Structuring database for multi-object "activity" and "following" functionalities

    - by romaninsh
    I am working on a web application which operate with different types of objects such as user, profiles, pages etc. All objects have unique object_id. When objects interact it may produce "activity", such as user posting on the page or profile. Activity may be related to multiple objects through their object_id. Users may also follow "objects" and they need to be able to see stream of relevant activity. Could you provide me with some data structure suggestions which would be efficient and scalable? My goal is to show activity limited to the objects which user is following I am not limited by relational databases. Update As I'm getting advices on ORM and how index things, I'd like to again, stress my question. According to my current design model the database structure looks like this: As you can see - it's quite easy to implement database like that. Activity and Follower tables do contain much larger amount of records than the upper level but it's tolerable. But when it comes for me to create a "timeline" table, it becomes a nightmare. For every user I need to reference all the object activities which he follows. In terms of records it easily gets out of control. Please suggest me how to change this structure to avoid timeline creation and also be abel to quickly retrieve activity for any given user. Thanks.

    Read the article

  • Browser-based GUI for a python application

    - by ack__
    I want to create a web/browser-based GUI for a command-line python application. The goal is to make use of HTML/JS technologies to create this GUI. As the application itself, it needs to run on Linux and Windows, and the interface will be accessible only from localhost (not exposed to internet). The GUI will contain 5 to 10 pages. I don't want a traditional desktop GUI that includes HTML/JS, but just a bunch of html files and some kind of controller between those and the application. I also want to make use of asynchronous programming (ajax like) so I can load and print data in the GUI without refreshing the whole page. I'd probably use jQuery for that and a couple other things. How would you recommend to design this? Performance is not the key here, I'm rather looking at reliability, portability and simplicity. I'm thinking of using a lightweight python HTTP server / framework (like CherryPy) and maybe later a Python templating system (at the begining it will just be a couple pages). EDIT: I'm looking for ideas/recommendations how to build this, not for alternatives to browser/web-based GUI.

    Read the article

  • Character progression through leveling, skills or items?

    - by Anton
    I'm working on a design for an RPG game, and I'm having some doubts about the skill and level system. I'm going for a more casual, explorative gaming experience and so thought about lowering the game complexity by simplifying character progression. But I'm having trouble deciding between the following: Progression through leveling, no complex skill progression, leveling increases base stats. Progression through skills, no leveling or base stat changes, skills progress through usage. Progression through items, more focus on stat-changing items, items confer skills, no leveling. However, I'm uncertain what the effects on gameplay might be in the end. So, my question is this: What would be the effects of choosing one of the above alternatives over the others? (Particularly with regards to the style and feel of the gameplay) My take on it is that the first sacrifices more frequent rewards and customization in favor of a simpler gameplay; the second sacrifices explicit customization and player control in favor of more frequent rewards and a somewhat simpler gameplay; while the third sacrifices inventory simplicity and a player metric in favor of player control, customization and progression simplicity. Addendum: I'm not really limiting myself to the above three, they are just the ones I liked most and am primarily interested in.

    Read the article

  • Designing a Content-Based ETL Process with .NET and SFDC

    - by Patrick
    As my firm makes the transition to using SFDC as our main operational system, we've spun together a couple of SFDC portals where we can post customer-specific documents to be viewed at will. As such, we've had the need for pseudo-ETL applications to be implemented that are able to extract metadata from the documents our analysts generate internally (most are industry-standard PDFs, XML, or MS Office formats) and place in networked "queue" folders. From there, our applications scoop of the queued documents and upload them to the appropriate SFDC CRM Content Library along with some select pieces of metadata. I've mostly used DbAmp to broker communication with SFDC (DbAmp is a Linked Server provider that allows you to use SQL conventions to interact with your SFDC Org data). I've been able to create [console] applications in C# that work pretty well, and they're usually structured something like this: static void Main() { // Load parameters from app.config. // Get documents from queue. var files = someInterface.GetFiles(someFilterOrRegexPattern); foreach (var file in files) { // Extract metadata from the file. // Validate some attributes of the file; add any validation errors to an in-memory // structure (e.g. List<ValidationErrors>). if (isValid) { var fileData = File.ReadAllBytes(file); // Upload using some wrapper for an ORM or DAL someInterface.Upload(fileData, meta.Param1, meta.Param2, ...); } else { // Bounce the file } } // Report any validation errors (via message bus or SMTP or some such). } And that's pretty much it. Most of the time I wrap all these operations in a "Worker" class that takes the needed interfaces as constructor parameters. This approach has worked reasonably well, but I just get this feeling in my gut that there's something awful about it and would love some feedback. Is writing an ETL process as a C# Console app a bad idea? I'm also wondering if there are some design patterns that would be useful in this scenario that I'm clearly overlooking. Thanks in advance!

    Read the article

  • Is there a pattern or best practice for passing a reference type to multiple classes vs a static class?

    - by Dave
    My .NET application creates HTML files, and as such, the structure looks like variable myData BuildHomePage() variable graph = new BuildGraphPage(myData) variable table = BuildTablePage(myData) BuildGraphPage and BuildTablePage both require access data, the myData object. In the above example, I've passed the myData object to 2 constructors. This is what I'm doing now, in my current project. The myData object, and it's properties are all readonly. The problem is, the number of pages which will require this object has grown. In the real project, there are currently 4, but the new spec is to have about 20. Passing this object to the constructor of each new object and assigning it to a field is a little time consuming, but not a hardship! This poses the question whether it's better practice to continue as I have, or to refactor and create a new static class for myData which can be referenced from any where in my project. I guess my abilities to use Google are poor, because I did try and find an appropriate pattern as I am sure this type of design must be common place but my results returned nothing. Is there a pattern which is suited, or do best practices lean towards one implementation over another.

    Read the article

  • What are the factors affecting a new programming language?

    - by Saurav Sengupta
    I am developing a new general-purpose programming language of my own design. It's currently my own personal project. I have read of some experts saying that new languages do not usually survive (unfortunately I can't find a reference to that right now). What are the most substantial problems that a new language faces? The language syntax is similar to C/Python families, it does not use S-expressions, and it is an imperative language, but I'm doing first-class functions in it to provide the facilities of currying. In particular, I am concentrating on translating the source language to an intermediate language for execution by an interpreter, but I'm not in a position to translate to native code yet. What would be the issues with that? I've not personally used many non-native code languages, so I'm not well aware of the performance issues on today's machines. I also can't decide upon a lexer and parser generator. What would be the pros and cons of Flex and Yacc vs. hand-made? And what benefits will LLVM provide? I need to get the interpreter ready as quickly as possible. Finally, what factors will affect the language's use post release? I am planning a small library of essentials and full documentation for the first phase.

    Read the article

  • Highly scalable and dynamic "rule-based" applications?

    - by Prof Plum
    For a large enterprise app, everyone knows that being able to adjust to change is one of the most important aspects of design. I use a rule-based approach a lot of the time to deal with changing business logic, with each rule being stored in a DB. This allows for easy changes to be made without diving into nasty details. Now since C# cannot Eval("foo(bar);") this is accomplished by using formatted strings stored in rows that are then processed in JavaScript at runtime. This works fine, however, it is less than elegant, and would not be the most enjoyable for anyone else to pick up on once it becomes legacy. Is there a more elegant solution to this? When you get into thousands of rules that change fairly frequently it becomes a real bear, but this cannot be that uncommon of a problem that someone has not thought of a better way to do this. Any suggestions? Is this current method defensible? What are the alternatives? Edit: Just to clarify, this is a large enterprise app, so no matter which solution works, there will be plenty of people constantly maintaining its rules and data (around 10). Also, The data changes frequently enough to say that some sort of centralized server system is basically a must.

    Read the article

  • Command-Query-Separation and multithreading safe interfaces

    - by Tobias Langner
    I like the command query separation pattern (from OOSC / Eiffel - basically you either return a value or you change the state of the class - but not both). This makes reasoning about the class easier and it is easier to write exception safe classes. Now, with multi threading, I run into a major problem: the separation of the query and the command basically invalidates the result from the query as anything can happen between those 2. So my question is: how do you handle command query separation in an multi-threaded environment? Clarification example: A stack with command query separation would have the following methods: push (command) pop (command - but does not return a value) top (query - returns the value) empty (query) The problem here is - I can get empty as status, but then I can not rely on top really retrieving an element since between the call of empty and the call of top, the stack might have been emptied. Same goes for pop & top. If I get an item using top, I can not be sure that the item that I pop is the same. This can be solved using external locks - but that's not exactly what I call threadsafe design.

    Read the article

  • What is a good basic/flexible cms for a small website? [closed]

    - by Samuel
    Possible Duplicate: Which Content Management System (CMS) should I use? I'm designing a very basic portfolio website for an artist. It features a blog, portfolio, cv and contact page. I've handcoded the basics of this site in php/java, as it is a very small website (and I like coding by hand). But I need a simple cms backend for the dynamic parts of the website (the blog/portfolio). The big systems (ruby, joomla, wordpress) are far too invasive for my liking (and frankly a bit beyond my capabilities). Wordpress for example, requires too much adaptation of the design to the wordpress structure, and ruby is far too extensive for a simple site like this (in my opinion). So what I'm looking for is a (preferably open source) cms that has a simple backend for the artist to use as a blogger, with a mysql database for the content, that will allow me to insert content with simple tags (using smarty tags for example), but is otherwise not too invasive or demanding in terms of the required page structure. Does anyone know of a good cms that fits this description? p.s.: I have tried phpnews and cmsmadesimple, but phpnews was a litte too basic (but very close too what I'm looking for) and cmsmadesimple was way too slow (but otherwise also pretty close too what I wanted, though a bit too extensive).

    Read the article

  • Intelligence as a vector quantity

    - by Senthil Kumaran
    I am reading this wonderful book called "Coders at Work: Reflections on the Craft of Programming" by Peter Seibel and I am at part wherein the conversation is with Joshua Bloch and I found this answer which is an important point for a programmer. The paragraph, goes something like this. There's this problem, which is, programming is so much of an intellectual meritocracy and often these people are the smartest people in the organization; therefore they figure they should be allowed to make all the decisions. But merely the fact they are the smartest people in the organization does not mean that they should be making all the decisions, because intelligence is not a scalar quantity; it's a vector quantity. Here at the last sentence, I fail to get the insight which is he trying to share. Can someone explain it in a little further as what he means by a vector quantity, possibly trying to present the same insight. Further down, I get the point that he is not taking about having an organization where non-technical people (sometimes clueless) can be managers of the technical people for some reason that they can spend more time to write emails well, because the very next statement following the above paragraph was. And if you lack empathy or emotional intelligence, then you shouldn't be designing APIs or GUIs or languages. I understand that he is saying that in Software engineering, programmers should know how the users will see their product and design for them. I felt the above paragraph was very interesting.

    Read the article

  • How was Git designed?

    - by Mark Canlas
    My workplace recently switched to Git and I've been loving (and hating!) it. I really do love it, and it is extremely powerful. The only part I hate is that sometimes it's too powerful (and maybe a bit terse/confusing). My question is... How was Git designed? Just using it for a short amount of time, you get the feel that it can handle many obscure workflows that other version control systems could not. But it also feels elegant underneath. And fast! This is no doubt in part to Linus's talent. But I'm wondering, was the overall design of git based off of something? I've read about BitKeeper but the accounts are scant on technical details. The compression, the graphs, getting rid of revision numbers, emphasizing branching, stashing, remotes... Where did it all come from? Linus really knocked this one out of the park and on pretty much the first try! It's quite good to use once you're past the learning curve.

    Read the article

  • Different methods of ammo resupply

    - by Chris Mantle
    I'm writing a small game at the moment. Presently, I have one or two design elements that aren't locked down yet, and I wanted to ask for input on one of these. For dramatic effect, the player's character in my game is immobilised, alone and has a supposedly limited amount of ammo for their weapons. However, I would like to periodically resupply the player with ammo (for the purpose of balancing the level of difficulty and to allow the player to continue if they're doing well). I'm trying to think of a method of resupply that's different to the more familiar strategies of making ammo magically appear or having the antagonists drop some when they die. I'd like to emphasise the notion of the player's isolation as much as possible, and finding a way of 'sneaking' ammo to the player without removing too much of that emphasis is basically what I'm trying to think of (it's definitely a valid argument that resupplying the player removes it anyway) I have considered a sort of simple in-game 'store', where kills get you points that you can spend on ammo for your favourite weapon. This might work well, and may also be good for supporting a simple micro-transaction business model within the game. However, you'd have to pause the game often to make purchases, which would interrupt the action, and it works against the notion of isolation. Any thoughts?

    Read the article

  • Auto update for application hosted on multiple servers on cloud

    - by mots_g
    I'm working on an application which will run on multiple Amazon EC2 instances. I wish to incorporate auto update feature for my application. The updater should update all the Ec2 instances. Also, there is a central server which governs the creation/termination of EC2 instances as per load. The central server creates a EC2 new instance from a pre-configured custom AMI (custom image which has our application pre-installed). Also, once there is an update, the pre-configured AMI needs to be updated too else it would create new instances which are not updated. Should the central server notify all the ec2 instances for an update and then the instances update themselves?Or should the application on Ec2 instance have a check for periodically updating themselves? Also how should the Amazon custom AMI be updated? Should a new instance be created from it, updated and then a new AMI be re-created and then new images be created from this AMI? What is the best way to incorporate an auto update feature for this architecture? The central server is written in Java and the application running on the cloud is written in C++. Is there a good framework available that can be used for this architecture? Please let me know on what I could be missing in the design and how it would help me to have a nice, extensible and fail safe auto update architecture. Thanks

    Read the article

  • Programming and Ubiquitous Language (DDD) in a non-English domain

    - by Sandor Drieënhuizen
    I know there are some questions already here that are closely related to this subject but none of them take Ubquitous Language as the starting point so I think that justifies this question. For those who don't know: Ubiquitous Language is the concept of defining a (both spoken and written) language that is equally used across developers and domain experts to avoid inconsistencies and miscommunication due to translation problems and misunderstanding. You will see the same terminology show up in code, conversations between any team member, functional specs and whatnot. So, what I was wondering about is how to deal with Ubiquitous Language in non-English domains. Personally, I strongly favor writing programming code in English completely, including comments but ofcourse excluding constants and resources. However, in a non-English domain, I'm forced to make a decision either to: Write code reflecting the Ubiquitous Language in the natural language of the domain. Translate the Ubiquitous Language to English and stop communicating in the natural language of the domain. Define a table that defines how the Ubiquitous Language translates to English. Here are some of my thoughts based on these options: 1) I have a strong aversion against mixed-language code, that is coding using type/member/variable names etc. that are non-English. Most programming languages 'breathe' English to a large extent and most of the technical literature, design pattern names etc. are in English as well. Therefore, in most cases there's just no way of writing code entirely in a non-English language so you end up with a mixed languages. 2) This will force the domain experts to start thinking and talking in the English equivalent of the UL, something that will probably not come naturally to them and therefore hinders communication significantly. 3) In this case, the developers communicate with the domain experts in their native language while the developers communicate with each other in English and most importantly, they write code using the English translation of the UL. I'm sure I don't want to go for the first option and I think option 3 is much better than option 2. What do you think? Am I missing other options?

    Read the article

< Previous Page | 105 106 107 108 109 110 111 112 113 114 115 116  | Next Page >