RE: cvs update; merge

[Paul Sander]

>--- Forwarded mail from address@hidden

>[ On Thursday, August 30, 2001 at 16:23:33 (-0700), Paul Sander wrote: ]
>> Subject: RE: cvs update; merge
>>
>> RCS is capable of storing arbitrary data types in its delta format.  The RCS
>> file format standard is also extensible and is capable of storing whatever
>> additional meta-data are needed to facilitate the new capability.  RCS lacks
>> the tool to create arbitrary newphrases, but that's not barrier for a tool
>> that already writes RCS files from scratch.

>While that might be true in some sense, it's not exactly true (RCS uses
>the standard unix line-based diff algorithm to calculate the deltas it
>stores), and in any case the result is most often pretty useless to
>anything but RCS.

RCS recommends using Gnu Diff, and by default it uses the option that
computes diffs even on files that don't look like text files.  This
has already proven for many years that it is capable of storing binary
files, and that doing so works well enough with CVS to make it viable
for basic checkout/commit activities.

I'm not sure what you mean about the result being useless to anything
but RCS, but two things occur to me:  RCS is really the only tool that
needs to scan the deltas of binary files, and tools that are able to
scan RCS files will continue to be able to scan them as long as the
standard is adhered to.

>The reason this is important is because one of the very cool things
>about RCS is that the archive format itself is a *TEXT* file.  This
>means that if it gets borked you can just 'vi foo,v' and fix it!  Try
>doing that once you've put a binary into it!

So you must use a binary file editor to fix an RCS file that contains
binary data.  This isn't rocket science, as any Emacs user can tell you.
And there are plenty of other readily available binary editors if you
don't like Emacs.  But if it's really that important to keep your RCS
files as full text files then don't store binary files in them.  I think
most users don't care because they don't muck about directly with the
RCS files anyway.

>--- End of forwarded message from address@hidden