Re: Question re: deleted tags

[Derek Price]

Larry Jones wrote:

> Derek Price writes:
>  
>
> > Well, "fixing" it this way would put us back in the position where 
> > checkouts and updates could sometimes report that tags don't exist when 
> > they do.  I think this is worse than not checking out any data when the 
> > tag doesn't exist in the project or sub-project being checked out (which 
> > can happen anyhow when the tag is correctly in val-tags).  I just fixed 
> > this on feature by having the tag command itself update the "cache".
> >    
>
> I think you've forgotten how val-tags works.  If a tag appears in the
> val-tags file, it is assumed to be valid and no error is issued, even
> though the tag may not appear in any of the files being processed.  If
> the tag does not appear in the val-tags file, then CVS examines all of
> the files to be processed.  If the tag occurs in any of them, then it is
> valid and is added to the val-tags file; if not, then -- and only then
> -- do you get the "tag does not exist" error.
>  

I think that may have been the intention, but I don't think that's the 
way it really works.

First, take a look at the dirs2-10 & rmadd-17 tests on stable (they've 
been eliminated on feature).  They show that commit commands and 
sometimes update commands will sometimes both issue "no such tag" errors 
when they shouldn't.  Hence the warning in the manual about trying "some 
combination of commands" until val-tags gets updated when encountering 
this message.  So, obviously, some commands were not recursing when they 
didn't find the tag in val-tags.

It's possible I took the wrong approach to fixing this bug.  Not that 
what I did was bad, but perhaps I should have figured out why some 
commands were failing to search for the tag and fixed those?

Derek