Re: How well does CVS handle other types of data?

[Lan Barnes]

"Greg A. Woods" wrote:
> 
> [ On Wednesday, July 11, 2001 at 11:18:01 (-0700), Lan Barnes wrote: ]
> > Subject: Re: How well does CVS handle other types of data?
> >
> > Hmm. If I exclude the binaries in my projects from CVS, how can I apply
> > tags to them so that I know which revisions of each binary are meant to go
> > with which releases?
> 
> Man oh man oh man!  If there ever was an FAQ about CVS this must be it!
> Does this have to be spelled out in some concrete script form every two
> weeks, or what?
> 
> The answer is to use naming conventions on your binary files (and/or the
> directories they are archived in) such that their relationship to your
> release management process can be determined by their name alone. 

Well, that's nice, but it's adding another unnecessary process, and one
that I doubt that I can impose. First of all, for some of our binaries, the
file names are imposed by the development tool. Secondly, we already have
naming conventions in several of our projects, and I would be overburdening
the names to add more meaning to them.

I really fail to see the technical reason for deprecating using CVS
archives to store -- and tag -- binaries necessary for a build. The disk
space involved is a nonissue for us (space is cheap), and the issue of
avoiding simultaneous edits exists under any system.

I could wish that this almost visceral disdain for binary source that some
in the CVS community seem to feel would soften a bit. Binary source is a
reality of many projects. I agree that a totally binary project might be
better served by some other tool. But if I'm dealing with an icon or .gif
that has changed from release to release, especially in size, then I damn
well want it to be archived with version labels.

-- 
Lan Barnes                 address@hidden
Icon Consulting, Inc       858-273-6677

"Learn Linux in 2 Days"
          - Cover line on Linux Summit '99 brochure