Re: Preventing mistakenly deleting branch tags

[Stephen Cameron]

--- Donald Sharp <sharpd@cisco.com> wrote:
regarding my patch about deleting branch tags.

> This doesn't stop people from pulling the trigger.  It just makes
> it annoying and the user will definately figure it out.
> 
[...]
Yeah, my intent was only to help prevent accidental deletion of branch tags.
If a misguided user is bent on deleting a branch tag, my patch wouldn't stop
them.  Extending your analogy, my patch is like a safety switch on a gun.

I'm not too interested in extending it to something complicated involving
figuring out what user is running, etc, simply because deleting a branch tag is
a pretty questionable action no matter what user is running.  The thing to do
would probably be make sure that all documentation explaining "cvs tag -d -b"
indicates what a horribly bad idea actually using it is, so that when the user
does figure it out, he also figures out he doesn't want to do it.

That, or forbid moving branch tags altogether, (except via "cvs admin -n"...I
think it can be used to delete any tag you care to delete.)

Hmm, it occurs to me that "cvs tag -F" which can move a tag should probably be
a bit wary of moving branch tags too.  I've had novice users in the past commit
files to a branch, then do something very very bad, like this:

cvs tag -F branch_tag foo.c   

thinking they needed to "promote" the new revision to get it into the build, or
whatever.  This converts "branch_tag" into a non-branch tag, and this can go
unnoticed for quite awhile because it doesn't really break things right away. 
Eventually, someone notices they can't commit foo.c on branch_tag anymore, or
whatever other problem surfaces, and you have to go back and manually fix it.

If I make "cvs tag -d" wary of branch tags, I probably should make "cvs tag -F"
wary as well.

-- steve



__________________________________________________
Do You Yahoo!?
Get personalized email addresses from Yahoo! Mail - only $35 
a year!  http://personal.mail.yahoo.com/