Skip to main content

We newbies and the application design and architecture

After being around for some time in this world of IT I have started to realize why these experienced guys are so furious with the newbies. I use to read a lot of newbie bashing on web and use to think WTF. Now, I must say "Old is Gold".


There are three types of newbies I have come across. One who really do not want to design and just wants to get the job done in a procedural way. Second who know a thing or two about design and want to design there application the best possible way (constraint being there knowledge). Third who thinks they know everything about design and that is how things should be designed.


The real problem arises when they are working on an application from scratch. The first type may come up with a class or two and finish the code in a monolithic manner. The third type will create a design either too good or to brittle. These two types will surely get the work done even though extensibility and maintainability may not be there and thus the cost of every change will be just too high. 


I have found the second type getting stuck more often then not. They are always circumspect about there design and some times they just think too much about it. They google for some hints and solutions (more often then not they find something). The problem is knowing whether the solution is right for there case or not. 


I am not against any one type and neither I favor one. I put them down just so that one can realize where they fall. I have been lucky to get chance to work on an application or module design from scratch and some time on existing applications as well. Initially I found really difficult to move even an inch as I was two confuse about my design. I have been the second type and I have been the third. 


Here are few of my learning's:



  1. Never be the first type. You never know you might be the one working on the change tomorrow. 
  2. Design evolves. There is know best design. You can always better it gradually. 
  3. Follow the basic principles. Separation of concerns, Coupling, Cohesiveness are the few principles which I think if followed will always allow you to refactor your code latter.
  4. Work in layers when needed. 
  5. You might be the best around (for the Third type) but may be you can learn from others mistake or can make them understand their mistake. Keep looking for the better solution or for the mistake you should not do yourself. 
  6. You might be technically very strong but you need to understand the needs of your customer then only you can design something good for the customer. You are not designing for yourself. You are designing for the customer. There is cost associated with whatever you do (do not spend too much at design) and obviously there will be cost associated with any change (do spend time on design).
  7. There are simple techniques which when followed always allows you to improve latter. e.g. interface based programming, DI etc. Do the small things right and the rest will follow.
  8. Do not hesitate in doing pair programming. You will always learn.



Most importantly do not just remain stuck. Start with something and realize the shortcomings and fix them. I am still learning and believe that we learn from our mistakes. I still believe I have spent too much time analyzing and trying to get the things right the first time (Being stuck). That said, please do invest lot of time at architectural design as module design can change comparatively easily and with less cost.

Labels:

Comments

Post a Comment

Popular posts from this blog

Laravel XAMPP MySQL artisan migrate install error mysql.sock

In my previous post I wrote about setting up Laravel 4 on Mac with XAMPP . When I tried to use migrations (using artisan) I faced some issue. I will also post about Laravel migrations soon.    php artisan migrate:install Error :                                                     [PDOException]                                       SQLSTATE[HY000] [2002] No such file or directory                                              Solution : We need to tell artisan which mysql we want it to use. For this to work we need to porivde it with mysql.sock which we want it to use. So change your database settings like this: 'mysql' => array( 'driver'    => 'mysql', 'host'      => 'localhost',     'unix_socket' => '/Applications/xampp/xamppfiles/var/mysql/mysql.sock', 'database'  => 'wedding', 'username'  => 'root', 'password'  => '', '
6 comments
Read more

Add (Drop) a new article (table) at publisher in merge replication

Let us add the article using the GUI first. First of all create the table on publisher database. Now right click on the publisher and select properties from the menu. Click on articles. Uncheck the - "Show only checked articles in the list" checkbox, in order to see the newly added table. Select the table as shown in the figure below. Press ok The article has now been added to the publisher We now need to recreate the snapshot Right click the publication and select – “View snapshot agent status”. Click start to regenerate snapshot. Now right click on the subscription (I have both on same server you may have on different servers) and select “View synchronization status” Click on start on the agent. The schema changes will propagate to the client if you have "Replicate schema changes" property set to true in the publisher.
Post a Comment
Read more

Check SQL Server Job status (State) using sp_help_job and xp_sqlagent_enum_jobs

This article is about checking the status of a SQL job. In our work place we have lot of SQL jobs. These jobs will run whole day and are business critical. They will load the data and generate extracts which will be used by business people. Thus, it becomes quite essential to support the system efficiently so that the job finishes in time and as desired. Also, while designing a new system sometimes we need to check the dependency of one job over another. In such scenario we need to check whether a particular job has finished or not. All this can be achieved in SQL Server by using the procedures:- sp_help_job xp_sqlagent_enum_jobs Note: xp_sqlagent_enum_jobs is an undocumented proc inside of sp_help_job and is used extensively to get SQL agent job information. sp_help_job: This procedure gives some insight into the status, and information, about a job. This stored procedure provides information such as last start time, job status etc. Syntax sp_help_job { [ @job_id= ] jo
2 comments
Read more