Wednesday, February 26, 2014

Software Version Controlling Best Practices

Always use version controlling
Use version controlling/source control even if one developer exists. This issue brings change tracking and versioning of the application continuously. For multi developers, it brings more benefits like co-working on same files and fast codebase updates.

Do not commit halfdone work

Each commit should consist of a complete task (issue, bugfix, etc.). Partial commits may cause unfunctionality, crash of application or even compilation errors.

Full Update -> Build -> Run -> Full Commit rule
For having no trouble, a stable version controlling process should be applied. According to this process, you should full update the code, then build and run your application. If application runs correctly, you should commit all of your updated files. While committing, inspect all of your changed files and changes. If all team members apply this process, application will always continue to run and no one will prevent others work. 

Commit atomicity/granularity
Do not use version controlling system to backup each single file separately and do not commit hundreds of files after days or weeks of development. Commits should be atomic, and as a best practice, should include files which are related with only one task. Late commits also brings more conflicts.

Use descriptive commit messages
Commit messages are important for tracking changes. Descriptive commit messages help developers to control old versions of code within a functional context, instead of date and time only.

Don't commit generated files
Use "ignore" funcionalities of your version controlling tool to ignore commiting and updating of generated files (e.g. target folders, user settings, IDE generated files, generated classes after each build etc.) By doing this, you will not waste your time for managing those unstable/unimportant files.

Continuous integration
Use continuous integration along with version controlling. By doing that, perform automated builds and deploys to a specific location. Also, configure the tool to create a separate version number (with a predefined versioning algorithm, e.g. <major_changes>.<minor_changes>.<bug_fixes>) for each build. This will be useful to trace tests and deployments.

Use qualified version controlling tools and plugins (especially for merge)
Version controlling tool quality is also important. It must at least show changed files clearly, give ability to merge conflicted files easily and view old versions of files. Besides, it will be better to have a branching function to manage versions and and issue tracking system synchronization to manage issue-related code file matchings.

Use branches for testing and different lines of code
You should use different code branches for development, testing and also production. If a version of code is being tested, development should not continue on that branch. Development can continue on another branch and when a new version will be tested after fixing some bugs and adding new properties, a new test branch should be provided with a new version. When a stable test branch is reached, it may be released as a production branch.

You can also take a look at the below link for version controlling basics:

Monday, April 22, 2013

A Theorical Look into Object/Resource Pool Pattern

Sometimes our software projects need costly object creations (database connections, socket connections, large graphical objects etc.). This cost may be about time and/or memory. Those objects may also be needed to be created frequently. At that moment, Object/Resource Pool pattern comes to help.

Object Pool pattern:
  • Reuses unused costly objects without re-creating those objects (e.g. books in a library).
  • May create objects eagerly before they are needed, to manage application performance.
  • May limit created number of objects to manage application performance and user access.

Working principle:
Object Pool Class Diagram
Client requests an object (Reusable) from object pool. Object pool has a list of predefined reusables and gives an available one from pool. For the client's point of view, given reusable is a new object but it is probably (object creation strategies will be told later) a pre-created object with a new object's field values. For the optimum performance, client should notify object pool when reusable is no longer needed. Just in case client doesn't notify, object pool may define a timeout for each created reusable. Object pool must also have a synchronization mechanism for reusable serving, especially on multithreaded or multiuser applications.

When to use/Trade offs:
If objects are not costly, object pool pattern should not be used because this pattern needs clients to inform object pool (brings extra code complexity) and object creation management code (synchronization of pool, object creation limitation strategy etc.). This extra management code also brings a little performance loss. If performance gain of reusable object usage is bigger than extra management code performance loss, and application must frequently create those reusable objects, object pool pattern is recommended to be used.

Implementation Strategies:
    Object pool limit:
  • If memory is limited and/or maximum number of clients are needed to be restricted, an object number limit value may be defined for object pool (limited pool).
  • If there is no restriction, limit value is not needed (unlimited pool).
  • If all object pool list is in use, new objects may be created (expanding pool) or client may be enforced to waiting (fixed pool).
   Eager/lazy creation:
  • If application startup is performed infrequently and startup time is not very important, object pool list objects may be created at the startup (eager ceration).
  • If no restrictions exist, no objects are needed to be created eagerly (lazy creation).
  • Some number of objects may be created eagerly, others may be created lazily according to the application parameters (mixed creation).
    Empty object pool strategy:
  • If object pool is expanding, create a new object and return it.
  • If object pool is not expanding, return null or enforce client to wait until an object is ready to be given.
    Synchronization strategy:
  • Object pool should have a synchronization mechanism for object serving. Otherwise, especially on multiuser systems, pooling system may fail.
    Unused object strategy:
  • Usage of the pool objects may be lower than expected in some cases. Object pool may have a pool object removing strategy (i.e. descreasing object limit) for increasing performance.
    Returning object strategy:
  • Clients should return pool objects back to the pool after their job is completed. But pool code can't control this and poor client code may return pool objects. For this situation, object pool may have a timer mechanism for given but unused pool objects.

According to the project requirements, some of those strategies or mixed versions should be selected and implemented for object pool. But "reusing available objects" principle is the key and can not be changed for any implementation strategy.

Example Implementations will exist on the next post.

Wednesday, February 20, 2013

Layers of a Standard Enterprise Application

In a standard enterprise application which has a database and graphical UI (web or desktop), there are some typical layers which constitute the application. In this article we will mention those layers and give some instructions about them. 

The general diagram is as below:

There are some important properties about that diagram: 
There are vertical and horizontal layers. Vertical layers may be thought as general application service libraries which can paralelly work on all horizontal layers and independent from each other. However, a horizontal layer's subject of interest is only its neighbour (top and bottom) horizontal layers, and they work sequentially from top-to-bottom and bottom-to-top. Lastly, user can only interact with topmost horizontal layer. 

Horizontal layers:

  • PresentationEvery enterprise application has a UI, in fact graphical UI. This UI can be web or desktop based, which doesn't matter. The rule is simple. UI takes user action and sends it to the controller. And at the end it shows result taken from controller to the user. UI can be implemented according to MVC, MVP, MVVM or another approach.
  • ControlHandles business logic of the application. Takes info from user and sends it to DB layer (DAO or ORM framework) and vice versa. Abstraction type of controller may vary (a separate control and business layers for example)  according to the application parameters or development patterns (MVC, MVP, MVVM, ...) but the main idea remains the same.
  • Data AccessDatabase handling layer of the application. It may contain entity definitions, ORM framework or DB connection codes having SQL sentences, according to the abstraction decision. Its role is getting data from controller, performing data operation on database and sending results again to controller (if result exists). Database independence is a very important plus for this layer, which brings flexibility.

Vertical layers:

  • SecuritySecurity is a general concept, which includes user authorization, user authentication, user role management, network or SQL injection attack prevention, data recovery etc but we grouped that items generally as "Security". Those issues must be handled for each horizontal layer if you want to have a fully secure software.
  • Logging: Logging is very important for maintenance and reporting in enterprise applications. An easily configurable, parameterized, readable, flexible and correctly implemented logging layer for all horizontal layers is very useful for both developers and users. 
  • i18ni18n (internationalization) brings multiple language support and user interface text changing capability without rebuilding code. It may also have localization property which can handle number, currency and date formatting. So i18n provides flexibility on applications and should not be ignored.
  • Exception Handling: Exception handling may also be included in "Security" or "Logging" layers. We should not use only general exception classes or absorb all exceptions if we want to develop a quality software. We sometimes need our special exception classes with their specific behaviours for better security, logging and application performance.

If you design an architecture which has that layer structure considering software quality factors (which are told here), you will probably have a successful enterprise software application.