Re: cvs tag: some slightly counterintuitive behaviour

[Todd Denniston]

Andy Jones wrote:
> 
> I've found some behaviour of cvs tag that breaks the principal of least 
> astonishment - at least, for me.
> 
> In short: cvs tag doesn't tag the files in the working directory, it tags the 
> files it thinks are there.
> 
> I'm preparing a release.  Some of the code is questionable - I don't know if 
> it is dead or not but it certainly does not want to be part of the release.  
> So I've been deleting these files from the working directory.
<SNIP>
1) `rm -f filename`
or
2) `cvs remove -f filename;cvs commit`
.......
1) tells the file-system to make it go away.

2) tells cvs to tell the file-system to make it go away & mark a new revision
on this branch as dead.

in case 1) as far as cvs is concerned it IS still there (go on, do a `cvs
update`, cvs will remind the file-system).

in case 2) for this branch cvs knows it is DEAD.

If you only want to tag SOME of the files then you need to list each one
explicitly on the tag command line.  I have done this in the cases where to
perform qualification testing we wanted some of the testing scripts not all,
because some of the scripts were the kind that in software told the box to
reset itself and get out of error conditions without power cycle or simply
were not expected to be used during test so don't provide them.
check the archive I might have dropped some scripts when I posted about only
tagging some files before.

--
Todd Denniston
Crane Division, Naval Surface Warfare Center (NSWC Crane) 
Harnessing the Power of Technology for the Warfighter