I have the benefit of a QA team where I work. teehee. Seriously, I have two distinct deployment environment. 1. J scripts are in a Database Table and loaded in sequence at run-time 2. J scripts are consolidated into one IJS file and distributed with the application.
For option #1, I try to logically group the script files by project. There are of course core or base scripts but each project gets its own loading sequence. The advantage of this is that there are no duplication of scripts and I found that I can easily narrow down the offending scripts during production errors this way. To debug this, I load a no profile J session, load each script one by one using a predefined sequence onto the session. Since most of the libraries have been "tested", I would just concentrate on the new script module/record coming from the database. For option #2, I compile everything into one big IJS file using the project manager. With this, I have to pay attention on when and what script files are added onto the compiled IJS file. What I normally do is create a main.ijs in the PM, go to the Library tab and select the library scripts to be added to the file after the stdlib.ijs. After this, its coding time. I now test the IJS file as a "whole" by loading a no profile J session and loading the IJS file and working from there. Doing either option, I would still create a Virtual Machine for the target (I actually have images that I just reuse) OS and test the code there prior to deploying for QA. Oh, regarding the comments, I find it helpful down the road that comments are included in the distribution files. I actually have a deployed script where the comments are two pages long and the actual J script is only 5 lines. :P -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Don Guinn Sent: Wednesday, July 21, 2010 6:40 AM To: Programming forum Subject: Re: [Jprogramming] library suggestion How do you developers test J? Surely you have a way to load J with all the comments and as separate script files rather than using the files available in the distribution. Why not make that available for those who want to see all the comments and multiple files loading as initialization. On Tue, Jul 20, 2010 at 1:50 PM, <[email protected]> wrote: > >Improved performance, if it exists, would certainly be > > ... nullified by the extra efforts one would have go through to to > make sure a fix to a borked line ends up in all[*] the other copies. > > Redundancy in code - just say no. Otherwise, what would come next? > Junk the definition of utility verbs because inline code has "improved > performance"? > > Martin > > [*]: Both in the packages you have on your system and those you are > going to install half a year later. > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
