Re: CVS

jba_at_mentor.dk (Jan B. Andersen) writes:

>We have been using CVS (on UNIX, System V) in about one year
>to control c and Pro*C code. We now like to use it in a project,
>where ORACLE's SQL*Forms, CASE*Dictionary and SQL*ReportWriter are used.
 

>Does anyone out there have experience in using CVS in such an
>environment?

No but I've used RCS and sccs and PVCS so I imagine CVS Aegis will provide similar functionality - in fact I've heard good things about CVS, can you confirm or deny ?

>The difficult task, I think, is to spot "the source", because
>most of "the source" is stored in the database?! So what should be put
>in the repository?

The source for a SQL*Forms is the .inp file, the .frm can be derived by running iag30 or generate.
With SQL*Menu it is a little bit difficult as the .dmm and the database have to be in synch, and you are restricted to only one set of system tables. Try unloading a .sql file from SQL*Menu (you can do that with genmenu50), putting that in CVS and creating the .dmm by loading in the .sql (via SQL*Plus) and running genmnenu50 to create the .dmm. It sounds tedious but a simple script will automate this. With SQL*Reportwriter it is even more complex as even the .rex file is not easy to put under CVS (it has fixed length records). The best thing to do is write a shell script to ensure that the headers that CVS generates are translated into fixed length records in the .rex file. Then follow the same procedure as SQL*Menu.

>Another problem is merging in automatically generated files from
>ex. SQL*Forms (*.inp -files)?! Is that a good idea ?

Sure, make can be easily made "Oracle-sensitive". Try putting something like .frm:.inp

   iag30 -to $1 /
or something similar.

Regards,
Martin Chesbrough
Principal Consultant
Oracle Systems Australia Received on Mon Nov 15 1993 - 11:47:44 CET