Re: Bug or known weakness in CVS RCS-keyword expansion?

[Peter Toft]

On Wed, 12 Jul 2006, Peter Toft wrote:

> On Wed, 12 Jul 2006, Stuart Cooper wrote:
> 
> > > try cvs update -A instead.
> > 
> > nope, that didn't work but it was well worth a try.
> 
> Yeah - I use "update -A" to get the current versions of all files, and yes 
> it clears the stickty tags. But as you write it is not a solution for what 
> I wrote yesterday.
> 
> > 
> > I've got a handscrawled note next to $Name$ in my reference guide which 
> > reads:
> > 
> > ################################################################
> > "This keyword is expanded only if one checks out with an explicit tag name"
> > eg cvs co -r tag gives a $Name: tag$ expansion of the keyword"
> > ################################################################
> > 
> > Obviously the rules are a bit more subtle than that. Keyword expansion 
> > happens
> > at checkout time, so I guess your update -r TAG1 causes a checkout of a.txt
> > only and not b.txt (regardless of whether it's update -A -r TAG1 or just
> > update -r TAG1) and the expansion of $Name$ happens to a.txt only as that's
> > the only file which gets checked out.
> 
> And I would actually prefer to have b.txt changed as well in my example.
> 
> > The Tag in the text file CVS/Entries is kept OK in each case:
> > TAG1
> > $ cat CVS/Entries
> > /a.txt/1.1.1.1/Tue Jul 11 23:59:34 2006//TTAG1
> > /b.txt/1.1.1.1/Wed Jul 12 00:14:22 2006//TTAG1
> > D
> > 
> > $ cat CVS/Entries
> > /a.txt/1.1.1.1/Tue Jul 11 23:59:34 2006//TTAG2
> > /b.txt/1.2/Wed Jul 12 00:14:16 2006//TTAG2
> > D
> > 
> > Notice when I do the cvs update -r TAG1 from a TAG2 state, I only get
> > one file updated:
> > $ cvs update -r TAG1 .
> > cvs update: Updating .
> > U b.txt
> > 
> > Only b.txt checked out by this revert-update, so only b.txt 
> > keyword-expanded.
> > 
> > I guess to be completely sure of $Name$ validity, you'd trash the
> > project completely
> > and do a fresh cvs co -r TAG1 project
> 
> Yep - which was a part of my example, but that is clearly a bad solution, 
> and I guess for most large projects NOT possible in reality.

Should I file a formal bug-request from my description to
http://savannah.nongnu.org/bugs/?? I really don't see that the current 
behaviour is logical nor desired.

Best

Peter Toft, Ph.D. address@hidden http://pto.linux.dk

"First they ignore you, then they ridicule you, then they fight you, then you 
win."
  --  Mahatma Gandhi