[WBEL-users] Bug tracking and source versioning software

[James Knowles]

As far as the Win32 desktop, TortoiseCVS rocks 
(http://www.tortoisecvs.org/). I've not used the SVN version 
(http://tortoisesvn.tigris.org/).

I have personal notes on-line that others have found useful. They've not 
been updated for a while, so YMMV. :-)
http://www.ifm-services.com/people/jamesk/papers/cms/cvs-win32-client.html

>CVS+Bugzilla. ... that they can work together, 
>

AFAIK, not out of the box. I know somebody wrote something that attempts 
to tie them together, but it was such a mess when I tried it I don't 
bother. Things may have improved since. [Anybody know better?]

I know that there's a loose coupler between P4 and BZ... but P4's not 
cheap. At least it has Linux clients. :-)

Also, don't forget ViewCVS, which is handy for browsing the repository 
via your favourite web browser.

>the one time I tried to install Bugzilla was a
>nightmare.
>  
>
BZ is a b***h to setup on WBEL or similar RH distro unless you don't 
include graphing. It's still a bit arduous, but straightforward.

The problem is that BZ needs v2.0+ of a particular lib -- gd IIRC -- 
which isn't in WBEL/RH, and my attempts to install in isolation failed. 
But we don't really need the graphing, so the rest installed fine.

I've not had real need to use svn, but the ability to bind multiple 
changes into an atomic change may cause us to switch. On the other hand, 
the per-file change logs that CVS imposes are nice. We have code that 
has multi-year change lists for each file because of this. It's nice for 
those once in a while moments when one needs to look back and see where 
a specific thing changed.

>Scarab 
>  
>
Not looked at it for quite some time, when it was an early alpha. I have 
mixed feelings about creating a link between a bug tracker and a RCS. 
Sure, they're related, but it's not a strong 1:1 correspondance unless 
you do coarse-grained checkins.

If I need to document changes that relate to a specific bug, I'm more 
inclined to tag the alterations with the bug number so they're easy to 
find via the web client. I tend to make a larger number of small 
incremental checkins to easily document multi-step changes via diff, 
rather than a massive checkin that's murder to decipher with diff.

I guess it depends a lot of your RCS and bug tracking philosophy.