.TH CVSQ-BRANCH 1 "April 24, 2007" "" ""
.SH NAME
cvsq-branch - make a failed commit work by sending it to a branch

.SH SYNOPSIS
\fBcvsq-branch\fR
.br

.SH DESCRIPTION
If a cvs commit scheduled by cvsq(1) fails due to a conflict, an
easy way to empty the rest of the queue is to send the commit in
question to a new branch and merge later. This is precisely what
this script does.

It assumes that last command that failed in the cvs queue was a
commit. It will try to bring the queue in a state that the commit
succeeds to a branch and the working directory will at the end be
restored and a simple update called. This allows the user to
immediately start with merging. It also will leave a file named
.cvsq-branch-<unique_identifier> to remind users in all working
directories of the merge still to be done. The tag identifying the
branch is cvsq-branch-<unique_identifier>, with the same identifier as
left in the working directory.

.SH NOTE
After calling \fBcvsq-branch\fR a call of \fBcvsq up\fR is needed to 
do the actual upload to the branch!

.SH MORE DETAILS
This section assumes some knowledge about the file structure in the
cvs queue, see cvsq-files(5).

After calling cvsq-branch, it is first checked that the todo-pre
queue is empty. The slot of the failed commit is the least slot
(different to 0) in the todo queue. The working directory is stored in
the dirnames directory for the appropriate slot.

Once the files are identified, a new slot is allocated and a copy of
the working directory is stored in <new-slot>-backup. Then a new
branch name is generated and in the actual working directory 
"cvs tag <tag>-start", "cvs tag -b <tag>" and "cvs update -r<tag>" 
are called. As the commits in the
queue refer to their versions via symbolic links to the CVS
directories in the working directory, this has the desired effect.

Finally in the new slot the following commands a scheduled.
.br
- move the working directory to ~/.cvsq-branch/backup/<tag>
.br
- copy the backup copy in the data directory to the old place of the
working directory
.br
- enter the restored working directory and call "cvs update"
.br
- create add and commit a file named .<tag>

In the post queue it is scheduled to remove the backup copy in the
data directory (not the one in ~/.cvsq-branch/backup, of course!).

.SH FILES
\fB~/.cvsq/ \fR
.br
The directory that cvsq stores the queue of the commands to
be executed. See cvsq-files(5) for a description of the file structure
and how to schedule commands other than by calling cvsq.

\fB~/.cvsq-branch/backup\fR
.br
A directory where in the end a copy of the unmodified working
directory will end up. This is useful, if the working directory
contained more changes than those scheduled for commit.

.SH SEE ALSO
.BR cvsq-merge "(1), "cvsq "(1), "cvs "(1), "cvsq-files "(5)"

.SH AUTHOR
Klaus Aehlig <aehlig@linta.de>