Search Results

Search found 39598 results on 1584 pages for 'good design'.

Page 15/1584 | < Previous Page | 11 12 13 14 15 16 17 18 19 20 21 22  | Next Page >

  • Would this be a good web application architecture?

    - by Gustav Bertram
    My problem Our MVC based framework does not allow us to cache only part of our output. Ideally we want to cahce static and semi-static bits, and run dynamic bits. In addition, we need to consider data caching that reacts to database changes. My idea The concept I came up with was to represent a page as a tree of XML fragment objects. (I say XML, but I mean XHTML). Some of the fragments are dynamic, and can pull their data directly from models or other sources, but most of the fragments are static scaffolding. If a subtree of fragments is completely static, then I imagine that they could unfold into pure XML that would then be cached as the text representation of their parent element. This process would ideally continue until we are left with a root element that contains all of the static XML, and has a couple of dynamic XML fragments that are resolved and attached to the relevant nodes of the XML tree just before the page is displayed. In addition to separating content into dynamic and static fragments, some fragments could be dynamic and cached. A simple expiry time which propagates up through the XML fragment tree would indicate that a specific fragment should periodically be refreshed. A newspaper section or front page does not need to be updated each second. Minutes or sometimes even longer is sufficient. Other fragments would be dynamic and uncached. Typically too many articles are viewed for them to be cached - the cache would overflow. Some individual articles may be cached if they are extremely popular. Functional notes The folding mechanism could be to be smart enough to judge when it would be more profitable to fold a dynamic cached fragment and propagate the expiry date to the parent fragment, or to keep it separate and simple attach to the XML tree when resolving the page. If some dynamic cached fragments are associated to database objects through mechanisms like a globally unique content id, then changes to the database could trigger changes to the output cache. If fragments store the identifiers of parent fragments, then they could trigger a refolding process that would then include the updated data. A set of pure XML with an ordered array of fragment objects (that each store the identifying information of the node to which they should be attached), can be resolved in a fairly simple way by walking the XML tree, and merging the data from the fragments. Because it is not necessary to parse and construct the entire tree in memory before attaching nodes, processing should be fairly fast. The identifiers of each fragment would be a combination of relevant identity data and the type of fragment object. Cached parent fragments would contain references to these identifiers, in order to then either pull them from the fragment cache, or to run their code. The controller's responsibility is reduced to making changes to the database, and telling the root XML fragment object to render itself. The Question My question has two parts: Is this a good design? Are there any obvious flaws I'm missing? Has somebody else thought of this before? References? Is there an existing alternative that I should consider? A cool templating engine maybe?

    Read the article

  • Is my class structure good enough?

    - by Rivten
    So I wanted to try out this challenge on reddit which is mostly about how you structure your data the best you can. I decided to challenge my C++ skills. Here's how I planned this. First, there's the Game class. It deals with time and is the only class main has access to. A game has a Forest. For now, this class does not have a lot of things, only a size and a Factory. Will be put in better use when it will come to SDL-stuff I guess A Factory is the thing that deals with the Game Objects (a.k.a. Trees, Lumberjack and Bears). It has a vector of all GameObjects and a queue of Events which will be managed at the end of one month. A GameObject is an abstract class which can be updated and which can notify the Event Listener The EventListener is a class which handles all the Events of a simulation. It can recieve events from a Game Object and notify the Factory if needed, the latter will manage correctly the event. So, the Tree, Lumberjack and Bear classes all inherits from GameObject. And Sapling and Elder Tree inherits from Tree. Finally, an Event is defined by an event_type enumeration (LUMBERJACK_MAWED, SAPPLING_EVOLUTION, ...) and an event_protagonists union (a GameObject or a pair of GameObject (who killed who ?)). I was quite happy at first with this because it seems quite logic and flexible. But I ended up questionning this structure. Here's why : I dislike the fact that a GameObject need to know about the Factory. Indeed, when a Bear moves somewhere, it needs to know if there's a Lumberjack ! Or it is the Factory which handles places and objects. It would be great if a GameObject could only interact with the EventListener... or maybe it's not that much of a big deal. Wouldn't it be better if I separate the Factory in three vectors ? One for each kind of GameObject. The idea would be to optimize research. If I'm looking do delete a dead lumberjack, I would only have to look in one shorter vector rather than a very long vector. Another problem arises when I want to know if there is any particular object in a given case because I have to look for all the gameObjects and see if they are at the given case. I would tend to think that the other idea would be to use a matrix but then the issue would be that I would have empty cases (and therefore unused space). I don't really know if Sapling and Elder Tree should inherit from Tree. Indeed, a Sapling is a Tree but what about its evolution ? Should I just delete the sapling and say to the factory to create a new Tree at the exact same place ? It doesn't seem natural to me to do so. How could I improve this ? Is the design of an Event quite good ? I've never used unions before in C++ but I didn't have any other ideas about what to use. Well, I hope I have been clear enough. Thank you for taking the time to help me !

    Read the article

  • Interfaces, Adapters, exposing business objects via WCF design

    - by Onam
    I know there have been countless discussions about this but I think this question is slightly different and may perhaps prompt a heated discussion (lets keep it friendly). The scene: I am developing a system as a means for me to learn various concepts and I came across a predicament which my brain is conflicting with. That is whether to keep my interfaces in a separate class library or should they live side by side my business objects. I want to expose certain objects via WCF, however refuse to expose them in its entirety. I am sure most will agree exposing properties such as IDs and other properties is not good practice but also I don't want to have my business objects decorated with attributes. The question: Essentially, I'll be having a separate interface for each of my objects that will essentially be exposed to the outside world (could end up being quite a few) so does it make sense to create a separate class library for interfaces? This also brings up the question of whether adapters should live in a separate class library too as ideally I want a mechanism from transferring from one object to the other and vice versa?

    Read the article

  • C# Design How to Elegantly wrap a DAL class

    - by guazz
    I have an application which uses MyGeneration's dOODads ORM to generate it's Data Access Layer. dOODad works by generating a persistance class for each table in the database. It works like so: // Load and Save Employees emps = new Employees(); if(emps.LoadByPrimaryKey(42)) { emps.LastName = "Just Got Married"; emps.Save(); } // Add a new record Employees emps = new Employees(); emps.AddNew(); emps.FirstName = "Mr."; emps.LastName = "dOOdad"; emps.Save(); // After save the identity column is already here for me. int i = emps.EmployeeID; // Dynamic Query - All Employees with 'A' in thier last name Employees emps = new Employees(); emps.Where.LastName.Value = "%A%"; emps.Where.LastName.Operator = WhereParameter.Operand.Like; emps.Query.Load(); For the above example(i.e. Employees DAL object) I would like to know what is the best method/technique to abstract some of the implementation details on my classes. I don't believe that an Employee class should have Employees(the DAL) specifics in its methods - or perhaps this is acceptable? Is it possible to implement some form of repository pattern? Bear in mind that this is a high volume, perfomacne critical application. Thanks, j

    Read the article

  • Domain driven design: Manager and service

    - by ryudice
    I'm creating some business logic in the application but I'm not sure how or where to encapsulate it, I've used the repository pattern for data access, I've seen some projects that use DDD that have some classes with the "Service" suffix and the "manager" suffix, what are each of this clases suppose to take care of in DDD?

    Read the article

  • Repeated properties design pattern

    - by Mark
    I have a DownloadManager class that manages multiple DownloadItem objects. Each DownloadItem has events like ProgressChanged and DownloadCompleted. Usually you want to use the same event handler for all download items, so it's a bit annoying to have to set the event handlers over and over again for each DownloadItem. Thus, I need to decide which pattern to use: Use one DownloadItem as a template and clone it as necessary var dm = DownloadManager(); var di = DownloadItem(); di.ProgressChanged += new DownloadProgressChangedEventHandler(di_ProgressChanged); di.DownloadCompleted += new DownloadProgressChangedEventHandler(di_DownloadCompleted); DownloadItem newDi; newDi = di.Clone(); newDi.Uri = "http://google.com"; dm.Enqueue(newDi); newDi = di.Clone(); newDi.Uri = "http://yahoo.com"; dm.Enqueue(newDi); Set the event handlers on the DownloadManager instead and have it copy the events over to each DownloadItem that is enqeued. var dm = DownloadManager(); dm.ProgressChanged += new DownloadProgressChangedEventHandler(di_ProgressChanged); dm.DownloadCompleted += new DownloadProgressChangedEventHandler(di_DownloadCompleted); dm.Enqueue(new DownloadItem("http://google.com")); dm.Enqueue(new DownloadItem("http://yahoo.com")); Or use some kind of factory var dm = DownloadManager(); var dif = DownloadItemFactory(); dif.ProgressChanged += new DownloadProgressChangedEventHandler(di_ProgressChanged); dif.DownloadCompleted += new DownloadProgressChangedEventHandler(di_DownloadCompleted); dm.Enqueue(dif.Create("http://google.com")); dm.Enqueue(dif.Create("http://yahoo.com")); What would you recommend?

    Read the article

  • RESTful Question/Answer design?

    - by Kirschstein
    This is a toy project I'm working on at the moment. My app contains questions with multiple choice answers. The question url is in the following format, with GET & POST mapping to different actions on the questions controller. GET: url.com/questions/:category/:difficulty => 'ask' POST: url.com/questions/:category/:difficulty => 'answer' I'm wondering if it's worth redesigning this into a RESTful style. I know I'd need to introduce answers as a resource, but I'm struggling to think of a url that would look natural for answering that question. Would a redesign be worthwhile? How would you go about structuring the urls?

    Read the article

  • Design Question on when to save

    - by Ben
    Hi, I was just after peoples opinion on when the best time to save an object (or collection of objects) is. I appreciate that it can be completely dependent on the situation that you are in but here is my situation. I have a collection of objects "MyCollection" in a grid. You can open each object "MyObject" in an editor dialogue by double clicking on the grid. Selecting "Cancel" on the dialogue will back out any changes you have made, but should selecting "ok" commit those changes back to the database, or should they commit the changes on that object back to the collection and have a save method that iterates through the collection and saves all changed objects? If i have an object "MyParentObject", that contains a collection of childen "MyChildObjectCollection", none of the changes made to each "MyChildObject" would be commited to the database until the "MyParentObject" was saved - this makes sense. However in my current situation, none of the objects in the collection are linked, therefore should the "Ok" on the dialogue commit the changes to the database? Appreciate any opinions on this. Thanks

    Read the article

  • User account design and security...

    - by espinet
    Before I begin, I am using Ruby on Rails and the Devise gem for user authentication. Hi, I was doing some research about account security and I found a blog post about the topic awhile ago but I can no longer find it again. I read something about when making a login system you should have 1 model for User, this contains a user's username, encrypted password, and email. You should also have a model for a user's Account. This contains everything else. A User has an Account. I don't know if I'm explaining this correctly since I haven't seen the blog post for several months and I lost my bookmark. Could someone explain how and why I should or shouldn't do this. My application deals with money so I need to cover my bases with security. Thanks.

    Read the article

  • Schema design: many to many plus additional one to many

    - by chrisj
    Hi, I have this scenario and I'm not sure exactly how it should be modeled in the database. The objects I'm trying to model are: teams, players, the team-player membership, and a list of fees due for each player on a given team. So, the fees depend on both the team and the player. So, my current approach is the following: **teams** id name **players** id name **team_players** id player_id team_id **team_player_fees** id team_players_id amount send_reminder_on Schema layout ERD In this schema, team_players is the junction table for teams and players. And the table team_player_fees has records that belong to records to the junction table. For example, playerA is on teamA and has the fees of $10 and $20 due in Aug and Feb. PlayerA is also on teamB and has the fees of $25 and $25 due in May and June. Each player/team combination can have a different set of fees. Questions: Are there better ways to handle such a scenario? Is there a term for this type of relationship? (so I can google it) Or know of any references with similar structures?

    Read the article

  • Project design / FS layout for large django projects

    - by rcreswick
    What is the best way to layout a large django project? The tutuorials provide simple instructions for setting up apps, models, and views, but there is less information about how apps and projects should be broken down, how much sharing is allowable/necessary between apps in a typical project (obviously that is largely dependent on the project) and how/where general templates should be kept. Does anyone have examples, suggestions, and explanations as to why a certain project layout is better than another? I am particularly interested in the incorporation of large numbers of unit tests (2-5x the size of the actual code base) and string externalization / templates.

    Read the article

  • General ORM design question

    - by Calvin
    Suppose you have 2 classes, Person and Rabbit. A person can do a number of things to a rabbit, s/he can either feed it, buy it and become its owner, or give it away. A rabbit can have none or at most 1 owner at a time. And if it is not fed for a while, it may die. Class Person { Void Feed(Rabbit r); Void Buy(Rabbit r); Void Giveaway(Person p, Rabbit r); Rabbit[] rabbits; } Class Rabbit { Bool IsAlive(); Person pwner; } There are a couple of observations from the domain model: Person and Rabbit can have references to each other Any actions on 1 object can also change the state of the other object Even if no explicit actions are invoked, there can still be a change of state in the objects (e.g. Rabbit can be starved to death, and that causes it to be removed from the Person.rabbits array) As DDD is concerned, I think the correct approach is to synchronize all calls that may change the states in the domain model. For instance, if a Person buys a Rabbit, s/he would need to acquire a lock in Person to make a change to the rabbits array AND also another lock in Rabbit to change its owner before releasing the first one. This would prevent a race condition where 2 Persons claim to be the owner of the little Rabbit. The other approach is to let the database to handle all these synchronizations. Who makes the first call wins, but then the DB needs to have some kind of business logics to figure out if it is a valid transaction (e.g. if a Rabbit already has an owner, it cannot change its owner unless the Person gives it away). There are both pros/cons in either approach, and I’d expect the “best” solution would be somewhere in-between. How would you do it in real life? What’s your take and experience? Also, is it a valid concern that there can be another race condition the domain model has committed its change but before it is fully committed in the database? And for the 3rd observation (i.e. state change due to time factor). How will you do it?

    Read the article

  • How to better design it ???

    - by Deepak
    public interface IBasePresenter { } public interface IJobViewPresenter : IBasePresenter { } public interface IActivityViewPresenter : IBasePresenter { } public class BaseView { public IBasePresenter Presenter { get; set; } } public class JobView : BaseView { public IJobViewPresenter JobViewPresenter { get { this.Presenter as IJobViewPresenter;} } } public class ActivityView : BaseView { public IActivityViewPresenter ActivityViewPresenter { get { this.Presenter as IActivityViewPresenter;} } } Lets assume that I need a IBasePresenter property on BaseView. Now this property is inherited by JobView and ActivityView but if I need reference to IJobViewPresenter object in these derived classes then I need to type cast IBasePresenter property to IJobViewPresenter or IActivityPresenter (which I want to avoid) or create JobViewPresenter and ActivityViewPresenter on derived classes (as shown above). I want to avoid type casting in derived classes and still have reference to IJobViewPresenter or IActivityViewPresenter and still have IBasePresenter in BaseView. Is there a way I can achieve it ?

    Read the article

  • Design issue with ATG CommercePipelineManager

    - by user1339772
    The definition of runProcess() method in PipelineManager is public PipelineResult runProcess(String pChainId, Object pParam) throws RunProcessException This gives me an impression that ANY object can be passed as the second param. However, ATG OOTB has PipelineManager component referring to CommercePipelineManager class which overrides the runProcess() method and downcast pParam to map and adds siteId to it. Basically, this enforces the client code to send only Map. Thus, if one needs to create a new pipeline chain, has to use map as data structure to pass on the data. Offcourse, one can always get around this by creating a new PipelineManager component, but I was just wondering the thought behind explicitly using map in CommercePipelineManager

    Read the article

  • Exploring the Factory Design Pattern

    - by asksuperuser
    There was an article here: http://msdn.microsoft.com/en-us/library/Ee817667%28pandp.10%29.aspx The first part of tut implemented this pattern with abstract classes. The second part shows an example with Interface class. But nothing in this article discusses why this pattern would rather use abstract or interface. So what explanation (advantages of one over the other) would you give ? Not in general but for this precise pattern.

    Read the article

  • Repository Design Pattern Guidance

    - by thefactor
    Let's say you have an MVVM CRM application. You have a number of customer objects in memory, through a repository. What would be the appropriate place to handle tasks that aren't associated with traditional MVVM tasks from a GUI? For example, let's say every few minutes you want to check to see if their address is valid and pop up a notification if it is not. Or you want to send out an hourly e-mail update. Or you want a window to pop up to remind you to call a customer at a specific time. Where does this logic go? It's not GUI/action-oriented, and it's not logic that would be appropriate for a repository, I think.

    Read the article

  • Domain-Driven-Design question

    - by Michael
    Hello everyone, I have a question about DDD. I'm building a application to learn DDD and I have a question about layering. I have an application that works like this: UI layer calls = Application Layer - Domain Layer - Database Here is a small example of how the code looks: //****************UI LAYER************************ //Uses Ioc to get the service from the factory. //This factory would be in the MyApp.Infrastructure.dll IImplementationFactory factory = new ImplementationFactory(); //Interface and implementation for Shopping Cart service would be in MyApp.ApplicationLayer.dll IShoppingCartService service = factory.GetImplementationFactory<IShoppingCartService>(); //This is the UI layer, //Calling into Application Layer //to get the shopping cart for a user. //Interface for IShoppingCart would be in MyApp.ApplicationLayer.dll //and implementation for IShoppingCart would be in MyApp.Model. IShoppingCart shoppingCart = service.GetShoppingCartByUserName(userName); //Show shopping cart information. //For example, items bought, price, taxes..etc ... //Pressed Purchase button, so even for when //button is pressed. //Uses Ioc to get the service from the factory again. IImplementationFactory factory = new ImplementationFactory(); IShoppingCartService service = factory.GetImplementationFactory<IShoppingCartService>(); service.Purchase(shoppingCart); //**********************Application Layer********************** public class ShoppingCartService : IShoppingCartService { public IShoppingCart GetShoppingCartByUserName(string userName) { //Uses Ioc to get the service from the factory. //This factory would be in the MyApp.Infrastructure.dll IImplementationFactory factory = new ImplementationFactory(); //Interface for repository would be in MyApp.Infrastructure.dll //but implementation would by in MyApp.Model.dll IShoppingCartRepository repository = factory.GetImplementationFactory<IShoppingCartRepository>(); IShoppingCart shoppingCart = repository.GetShoppingCartByUserName(username); //Do shopping cart logic like calculating taxes and stuff //I would put these in services but not sure? ... return shoppingCart; } public void Purchase(IShoppingCart shoppingCart) { //Do Purchase logic and calling out to repository ... } } I've seem to put most of my business rules in services rather than the models and I'm not sure if this is correct? Also, i'm not completely sure if I have the laying correct? Do I have the right pieces in the correct place? Also should my models leave my domain model? In general I'm I doing this correct according DDD? Thanks!

    Read the article

  • Is there a design pattern for this ?

    - by ytrewq
    I have a component that needs to call a specific service depending on the input it receives. So my component has to look at the input and based on a configuration that says "for this input call this service with this data" needs to call the proper service. The services have a common signature method and a specific one (each). I thought about an abstract class that includes the signatures for all three methods. The implementation for the two services will override all three methods (throwing NotImplementedException for the methods that are not supported by current service). A component that could be initialized with a map (that for each input type will have the type of the service to be called) will also be defined. Do you have a better approach to cope this scenario ?

    Read the article

  • Looking for an appropriate design pattern

    - by user1066015
    I have a game that tracks user stats after every match, such as how far they travelled, how many times they attacked, how far they fell, etc, and my current implementations looks somewhat as follows (simplified version): Class Player{ int id; public Player(){ int id = Math.random()*100000; PlayerData.players.put(id,new PlayerData()); } public void jump(){ //Logic to make the user jump //... //call the playerManager PlayerManager.jump(this); } public void attack(Player target){ //logic to attack the player //... //call the player manager PlayerManager.attack(this,target); } } Class PlayerData{ public static HashMap<int, PlayerData> players = new HashMap<int,PlayerData>(); int id; int timesJumped; int timesAttacked; } public void incrementJumped(){ timesJumped++; } public void incrementAttacked(){ timesAttacked++; } } Class PlayerManager{ public static void jump(Player player){ players.get(player.getId()).incrementJumped(); } public void incrementAttacked(Player player, Player target){ players.get(player.getId()).incrementAttacked(); } } So I have a PlayerData class which holds all of the statistics, and brings it out of the player class because it isn't part of the player logic. Then I have PlayerManager, which would be on the server, and that controls the interactions between players (a lot of the logic that does that is excluded so I could keep this simple). I put the calls to the PlayerData class in the Manager class because sometimes you have to do certain checks between players, for instance if the attack actually hits, then you increment "attackHits". The main problem (in my opinion, correct me if I'm wrong) is that this is not very extensible. I will have to touch the PlayerData class if I want to keep track of a new stat, by adding methods and fields, and then I have to potentially add more methods to my PlayerManager, so it isn't very modulized. If there is an improvement to this that you would recommend, I would be very appreciative. Thanks.

    Read the article

  • How to program for constraints/rules

    - by Gaurav
    First the background, during interviews in the past, many times I have been asked to design some or other variation of card game as programming puzzle, and I have tried to design it in OO way, but I have never been satisfied with my solutions. However it was not until recently that I realized that I had been approaching the problem from the wrong direction. Specifically I was trying to solve the problem by modeling individual card as an object. Problem with this is individual cards don't have any non-trivial intrinsic behavior and therefore are not suitable (or primary) candidate as objects. What is interesting and important about cards are rules and constraints, such as there could be only four suits, or only thirteen cards in each suit. Of course, then there are any number of rules for games. So my questions are Are there any idioms/constructs/patterns to program for rules & constraints. How many in 1 can be applied in conjunction with OO paradigm.

    Read the article

  • Advice on designing a robust program to handle a large library of meta-information & programs

    - by Sam Bryant
    So this might be overly vague, but here it is anyway I'm not really looking for a specific answer, but rather general design principles or direction towards resources that deal with problems like this. It's one of my first large-scale applications, and I would like to do it right. Brief Explanation My basic problem is that I have to write an application that handles a large library of meta-data, can easily modify the meta-data on-the-fly, is robust with respect to crashing, and is very efficient. (Sorta like the design parameters of iTunes, although sometimes iTunes performs more poorly than I would like). If you don't want to read the details, you can skip the rest Long Explanation Specifically I am writing a program that creates a library of image files and meta-data about these files. There is a list of tags that may or may not apply to each image. The program needs to be able to add new images, new tags, assign tags to images, and detect duplicate images, all while operating. The program contains an image Viewer which has tagging operations. The idea is that if a given image A is viewed while the library has tags T1, T2, and T3, then that image will have boolean flags for each of those tags (depending on whether the user tagged that image while it was open in the Viewer). However, prior to being viewed in the Viewer, image A would have no value for tags T1, T2, and T3. Instead it would have a "dirty" flag indicating that it is unknown whether or not A has these tags or not. The program can introduce new tags at any time (which would automatically set all images to "dirty" with respect to this new tag) This program must be fast. It must be easily able to pull up a list of images with or without a certain tag as well as images which are "dirty" with respect to a tag. It has to be crash-safe, in that if it suddenly crashes, all of the tagging information done in that session is not lost (though perhaps it's okay to loose some of it) Finally, it has to work with a lot of images (10,000) I am a fairly experienced programmer, but I have never tried to write a program with such demanding needs and I have never worked with databases. With respect to the meta-data storage, there seem to be a few design choices: Choice 1: Invidual meta-data vs centralized meta-data Individual Meta-Data: have a separate meta-data file for each image. This way, as soon as you change the meta-data for an image, it can be written to the hard disk, without having to rewrite the information for all of the other images. Centralized Meta-Data: Have a single file to hold the meta-data for every file. This would probably require meta-data writes in intervals as opposed to after every change. The benefit here is that you could keep a centralized list of all images with a given tag, ect, making the task of pulling up all images with a given tag very efficient

    Read the article

  • Why use link classes in oql instead of classes that contain links

    - by Isaac
    itop abstracts its very complex database design with an object query language (oql). For this there are classes definded, like 'Ticket' and 'Server'. Now a Ticket usually is linked to a Server. In my naive way I would give the Ticket class an attribute 'affected_server_list', where I could reference the affected servers. itop does it different: neither Servers nor Tickets know of each other. Instead there is a class 'linkTicketToServer', which provides the link between the two. The first thing I noticed is that it makes oql queries more complex. So I wondered why they designed it this way. One thing that occured to me is that it allows for more flexiblity, in that I can add links without modifying the original classes. Is this allready why one would implement it this way, or are there other reasons for this kind of design?

    Read the article

  • Putting altered social media logo icons on my website, can I get sued?

    - by Håkan Bylund
    I would say most websites with a somewhat thought-through graphical design use social media icons (i.e twitter, facebook, youtube, et.c) which are altered to fit the theme and design of the site. Now, my boss insist we only use the ones provided by say facebook or twitter themselfes (in fear of getting sued or lose credability), but sometimes it just doesnt look very good on the site. What is the common practice for these things? What do you risk by using an altered logo? What should I tell my boss? I'll provide a few examples, what'd happen if I put any of these on a site?

    Read the article

  • Whatchamacallit: You know how there are breadcrumbs and sliders and whatnot

    - by Richard
    What do you call it when a web site (especially corporate/retail) has a series of rows with thumbnails, each with a little caption/description beneath explaining some benefit or feature of a product or service. Is there a name for this? I'm building a theme that incorporates this kind of design and I was hoping there is some kind of shorthand for this design feature. If you don't know what I'm talking about, check out one of the links below. http://themeforest.net/item/revolution-minimalist-business-html-template/full_screen_preview/2295335 http://themes.two2twelve.com/preview?theme=freshserve

    Read the article

  • Exposing warnings\errors from data objects (that are also list returned)

    - by Oren Schwartz
    I'm exposing Data objects via service oriented assembly (which on future usages might become a WCF service). The data object is tree designed, as well as formed from allot of properties.Moreover, some services return one objects, others retrieve a list of them (thus disables throwing exceptions). I now want to expose data flow warnings and wondering what's the best way to do it having to things to consider: (1) seperation (2) ease of access. On the one hand, i want the UI team to be able to access a fields warnings (or errors) without having them mapping the field names to an external source but on the other hand, i don't want the warnings "hanged" on the object itself (as i don't see it a correct design). I tought of creating a new type of wrapper for each field, that'll expose events and they'll have to register the one's they care about (but totally not sure) I'll be happy to hear your thoughts. Could you please direct me to a respectful design pattern ? what dp will do best here ? Thank you very much!

    Read the article

< Previous Page | 11 12 13 14 15 16 17 18 19 20 21 22  | Next Page >