Re: What's remembered? "S" flag suggestion

[Eric Siegerman]

On Tue, Aug 07, 2001 at 10:33:05AM -0700, Kent Henneuse wrote:
> Eric Siegerman wrote:
> > If there's a sticky release tag on a file, cvs should tell me.
> > How about a new status code from "update":
> >         S foo.c
> 
> This would be fantastic.  I would think that cvs would only 
> show you sticky revisions and not things like branches.

Agreed.

> For the
> branch case cvs could just print out a line at the begining or
> better yet the end of a cvs update saying "In branch tag XYZ".

Hadn't thought of that.  Might be nice.  It should probably go to
stderr, and -q should suppress it.

> Don't know about the data structures.  How does cvs currently
> hold a file at a specified revision?  Doesn't seem like it 
> would be more work to just have it report it being sticky when
> it does the update.

True, for the simple version.  As you say, CVS already remembers
the revision at which my sandbox is stuck (in the
sticky-tag-or-date field of CVS/Entries); this is what's
displayed by "cvs status".

The difficulty is with (what I believe to be) the ideal version
of this feature -- only turning on the reports once a new
revision has been committed.  To implement that, CVS would also
need to remember which revision *was* most recent *at the time I
did the sticky update*.  So if, while 1.6 of foo.c was HEAD, I
did:
        cvs update -r1.5 foo.c
CVS would need to store *both* revision numbers 1.5 and 1.6.  It
currently stores the former, but not the latter.

--

|  | /\
|-_|/  >   Eric Siegerman, Toronto, Ont.        address@hidden
|  |  /
With sufficient thrust, pigs fly just fine. However, this is not
necessarily a good idea.
        - RFC 1925 (quoting an unnamed source)