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

[Jeff King]

I think there's a major difference between datatypes that are occasionally
difficult to automatically merge (source code) and datatypes that are
fundamentally unmergable (bitmaps).

> -----Original Message-----
> From: address@hidden [mailto:address@hidden Behalf Of
> Noel L Yap
> Sent: Wednesday, July 11, 2001 3:28 PM
> To: address@hidden
> Cc: address@hidden
> Subject: Re: How well does CVS handle other types of data?
>
>
> Not only that, but CVS does't handle code reformats elegantly at all.
> Philosophically, this means that CVS can't merge source code under all
> conditions.  Does this mean you shouldn't use it when such tasks
> can occur?  Of
> course not,  All it means is that these situations need to be
> controlled so that
> CVS is usable (eg minimize their occurrence and ensure others
> know that it's
> going on).  Even if the situation is cotrolled, there's still a
> chance that you
> have a massive manual merge to do (eg when branches are
> involved).  Does this
> mean that one should never do code reformats?
>
> I guess this is the real argument -- it's not whether the files
> are text or
> binary, it's whether they're mergable or not.  Ideally (and this has been
> discussed before), CVS would allow hooks for diff/merge engines.
> You could then
> supply such an engine for, say, Java code, and refomat as you desire.