Messy CVS

[Paul J. T. <th...@ok...>]

It has come to my attention that our cvs repository is getting a little bit, 
shall we say, messy. Particularly, the following issues have come to my mind.


1. Naming comventions for tags/branches
=======================================

This is not a big issue and is, you could say, mostly resolved. We had the tag 
for release 1.0.2 of SM named 'sm1-0-2', then we came along and created the 
branch for SM 1.1 and named it 'sm1_1'. Now, I now this isn't a REALLY big 
deal, but I want a good way to keep our branches and releases seperate, as well 
as have them all named consistently.

So, from now on, the following rules should be used concerned tags/branches:

  a. Whenever a release is made (ie 1.0.3, etc) the cvs files released
     should be tagged as 'rel-1_0_3' (where 1_0_3 == 1.0.3)
     (Sorry, but cvs does not allow . to be used in tag/branch names)

  b. As to follow the same naming convention, branches of squirrelmail
     should be named similarly, except prefixed with 'sm' instead of 'rel'.
     For instance, the SM 1.1 branch should be named sm-1_1.
     NOTE that this is different then what it currently is, sm1_1, and I
     will change it really soon. Just thought I would let people know first.


2. Branch versus the main trunk
===============================

I have been doing a bunch of reading in the CVS manual, documentation, etc. I 
believe that the way we have our SM1.0/SM1.1 stuff set up is wrong.

Currently, in the squirrelmail module in our cvs repository:
    Main Trunk    = SquirrelMail 1.0
    Branch sm-1_1 = SquirrelMail 1.1

I believe, from what I have read, it is a much better idea for the currently 
being developed code to be the main trunk, as follows:
    Main Trunk    = SquirrelMail 1.1
    Branck sm-1_0 = SquirrelMail 1.0

This is a small detail, however, I believe it will result in a lot cleaner CVS 
repository, in the long run. I have experimented around with a local 
installation and have figured out how to covert what we currently have into 
what I have suggested. However, since this is a bigger change then just 
renaming sm1_1 to sm-1_1, I wanted to get the opinions of the people on this 
list, first.

I feel very strongly about this. As in, what we are doing right now is the 
wrong way. It will take a LITTLE bit of work to fix right now. Later, it will 
either take much more, or will be impossible.

So, let me know what you think. I especially look forward to hearing the 
opinions of those who having used CVS quite a bit.


3. Where have all the Plugins gone...
=====================================

Isn't that a song from the sixties? Humm, I was born in 77, so I guess I 
wouldn't know.

Anyways, I am of the opinion that the plugins should be part of the main 
squirrelmail module in CVS. This would mean that they (like the rest of the 
squirrelmail module) would exist as both a sm-1.0 and sm-1.1 version.

I believe this because:

  a. They really are a part of SM core, just optional functionality.
  b. They are easier to work on when they are in the main module
  c. We WILL develop problems when something in SM1.1 changes that
     effects plugins, but the same change does not apply to SM1.0.
     When that happens, we will have to create two versions (sm1.0
     sm1.1) for all plugins affected, so I say we might as well go
     for it now.

For those that may be concerned, it is very easy to get CVS to merge changes 
from one branch of a module into another. This will help in maintaining plugins 
across both versions.

Anyways, let's talk this one over, too.


Enough for now! Good work, everyone...

-- 
Paul Joseph Thompson
Oklahoma State University
th...@ok...