[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Should commit really fall back to CVSROOT?

From: Doug Lee
Subject: Should commit really fall back to CVSROOT?
Date: Sat, 11 Nov 2006 08:02:25 -0500
User-agent: Mutt/1.5.9i

I have a sandbox checked out from one repository, but my default
CVSROOT points to another repository.  The local repository happens to
be on a removable drive (a thumb drive actually).  Now and then, I'll
issue a `cvs commit' from the sandbox but forget that my removable
drive is not plugged in.  The result is first an error in each
directory of the sandbox about the repository not existing, but then
an attempt to commit to the repository pointed to by my CVSROOT.  It
just so happens that, on the latest occasion, there was such a
repository there--it was a backup of the removable-drive repository.
The upshot was a commit got sent straight into the backup copy before
I realized what was happening.

My question:  Why, after the initial attempt to commit fails, is an
additional attempt made at all?  The sandbox contains specifications
as to where each directory of files came from.  I would think no
attempt should be made (at least without some specific direction from
the committer) to send things anywhere else.

I use cvs v1.11.17.

I ask this question more on principal than out of urgent need.  I
fixed the resulting inconsistency thus:

(back up sandbox files I had changed and wanted to commit)
cvs update  (trashed my changes on returning to the local repo copy)
(restore backups to regain my changes)
cvs commit ..

Of course this tactic only worked because no one else had made changes
I need to `update' into my sandbox...

Doug Lee                 address@hidden        
SSB + BART Group         address@hidden
"I am a leader by default, only because nature does not allow a vacuum."
Bishop Desmond Tutu

reply via email to

[Prev in Thread] Current Thread [Next in Thread]