Re: cvs server with WSAD -- JAVA400-L

Joe Pluta wrote:

I'm currently doing just that, using CVSNT for my server.  I posted a
question on IBM's mailing list but didn't get squat - maybe you can help?


Agree with DM that using CVSNT leaves you with restricted functionality
however if you are a linux newbie then it doesn't have the learning
curve that setting up CVS on Linux would have.

I'm trying to determine the best way (well, the easiest) to transfer all the
information in my VAJ repository to a CVS repository.  I have several
versions of code, and I'm trying to keep that version information.
To date, the best I've been able to theorize (and I haven't even gotten this
to work in practice yet) is to set up a repository in CVS for each project,
then repetitively check everything out into work folders, export the next
version from VAJ into the same work folders, then check that back into CVS.


I have never worked with VAJ so I am not sure how you would export all
the version histories with your projects. The CVS manual has some
references to other version control systems here ->
http://www.cvshome.org/docs/manual/cvs_3.html#SEC41 Ultimately I suspect
that if VAJ does not export to a common format then someone somewhere
has had the same problem and has created a tool for the job.

I'm still a complete neophyte, though, because I don't know whether I should
use separate repositories, or simply separate modules for each of my VAJ
projects.  My guess is that it really depends on how closely related the
projects are.  For example, I have four closely related projects (internal
classes, external interface, and a couple of test environments).  Those seem
to make sense as separate modules within a single repository.


A repository might be viewed more like a CVS environment. As mark says,
it is used to control access etc. In practice when and when not to use
more than one repository comes down to almost abitrary decisions like
Personal repository vs. Company repository, Team a vs team b, Lan
segment a vs. lan segment b.

More on multiple repositories here ->

http://www.cvshome.org/docs/manual/cvs_2.html#SEC22

Your example given sounds like a perfectly acceptable use of modules
although unless your test environments have different source code in
them from the core project (in which case they should probably be
branches in the same module ->
http://www.cvshome.org/docs/manual/cvs_5.html#SEC55 ) there is probably
no need for them to be separate modules. With CVS you checkout to
sandbox or multiple sandboxes and all testing can be done locally from
there. Maybe I have misunderstood 'test environments' since I am a
newbie Java kiddie.

I don't know anything about module lists; it looks like I have to manually
create them for each module.  Are module lists suggested?  Required?  Not
required but highly recommended?


You can actually checkout and work on files in the CVSHOME directory
(the dir that contains all the admin files). The modules file is not
required but is useful for browsing module lists etc. from the client.
You need to add module entries manually. More information can be found
here -> http://www.cvshome.org/docs/manual/cvs_3.html#SEC43 and also
here -> http://www.cvshome.org/docs/manual/cvs_2.html#SEC20

It is worth chucking EVERYTHING into CVS - we have moved all our admin
documents, documentation etc. The benefits have been enormous. Using
something like TortoiseCVS http://www.tortoisecvs.org/download.shtml
allows you to hook CVS commandline + client into Windows explorer.

Also available of the CVS site is the FAQ-o-Matic ->
http://ccvs.cvshome.org/fom//cache/1.html

--
Christopher Joseph

-------------------
http://www.ideadesigners.com  [iseries & web technologies]
mailto: christopher@xxxxxxxxxxxxxxxxx
ICQ: 78019724