httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From (Ralf S. Engelschall)
Subject [PAPER] Working with the Apache repository remotely
Date Wed, 22 Oct 1997 12:39:13 GMT

As Rob said, anyone interested can play with CVSup, so I did ;_) I already
used CVSup in the past for my FreeBSD work. So I decided to push CVSup for
Apache, too.  I've configured some more CVSup containers on taz's CVSupd, the
same way we have CVSupd configured at (for instance
cvs-all for the complete repository).

Altough Rob said that there is rare need for complete repository copies for us
developers, I wanted to try myself. The result impressed me:

1. With CVSup I was able to transfer the complete
   20 MB from taz:/home/cvs/ to my in less then one hour
   over my 64kbps link because of CVSups compression.

2. CVSup is really fast even for complete ,v file copies,
   so syncing a complete repository is really useful.  This works because
   CVSup transfers only diffs for ,v files.

3. I really want to use both repositories (local and remote)
   in parallel. To do this I wrote a little program (named sCVS) which lets me
   interactively select the repository and then switches all CVS/Root and
   CVS/Repository files. This way I can use "cvs update" via the local
   repository copy and "cvs commit" via the remote repository master.

Because one needs to setup a few not totally trivial things, I've
also written a paper today which explains the steps needed for
  - using direct remote CVS via over SSH
  - syncing a local CVS repository with remote one via CVSup
  - using a local and remote repository in parallel on
    the same working directory tree
Ok, at the end of the paper I've written down a few problems for the switching
part. Here I've to investigate more, when I find time. But the rest already
works fine. Give it a read-try:

Perhaps someone (Rob?) really tries what I've written down and gives
feedback... Very welcome are hints for the Problems at the end of the

                                       Ralf S. Engelschall

View raw message