CVS Usage on the GNU Web Site

OBSOLETED BY THE INTEGRATION OF www.gnu.org:~www/html/ INTO CVS.

Table of contents

• Areas of www.gnu.org Under CVS
• How To Put a Tree Under CVS

Areas of www.gnu.org Under CVS

(Note that some areas on this server, www.gnu.org, that are under CVS, have their CVSROOT on remote CVS servers.)

This page lists (most, if not all) the areas on www.gnu.org that are under CVS and how to add new areas to CVS. [@@ The instructions for adding new areas is not yet complete. --ttn, 2000/07/24 10:49:51]

We don't list volunteers who maintain these areas, because they might not want their email address exposed to spammers. Ask [email protected] to add them to ~www/GNU.web-folk.html which is not published on this site for that reason.

Brave GNU World

/home/www/html/brave-gnu-world/

GNU people home pages

greve

adns

/home/www/html/software/adns/

awacs

/home/www/html/software/awacs/

Guile

/home/www/html/software/guile/

Hurd

/home/www/html/software/hurd/

libtool

/home/www/html/software/libtool/

userv

/home/www/html/software/userv/

Xlogmaster

/home/www/html/software/xlogmaster/

• check CVSROOT/loginfo to see which areas of the web tree are automatically updated whenever a change is committed.
• check for CVS directories in each directory under /home/www/html/.

How To Put a Tree Under CVS

Before you put any of the web site under CVS please consult with [email protected].

Consult the CVS Manual as necessary.

If you want to put a portion of the web site under CVS, here's how to go about it:

1. Tell CVS you want to work with the GNU web site's CVS repository:

   $ export CVSROOT=/home/www/cvs

2. First of all, get the html module to look the way you want. This is the CVS module that contains all the web pages under CVS for www.gnu.org. At the end of this step, your subtree under the "live" tree (/home/www/html/SUBTREE) will be checked into CVS as html/SUBTREE.
   • If you are adding a subtree that already exists on the web server to CVS for the first time, and you do not have any �Emacs-style (.~N~) backup files, then say:

     	      $ cd /home/www/html/SUBTREE
     	      $ cvs import html/SUBTREE $USER first
     	      

     CVS will start up an editor and prompt you for a log message. Enter the message, write it, exit the editor, and you are done with this step.

   • If you are adding a subtree that already exists on the web server to CVS for the first time, and you have Emacs-style (.~N~) backup files that you want to save as CVS revisions (this is the preferred default on this server, www.gnu.org), then for each directory DIR in SUBTREE, do:

     	      $ cd /home/www/html/DIR
     	      $ mkdir /home/www/cvs/html/DIR
     	      $ chmod 775 /home/www/cvs/html/DIR
     	      $ chmod g+s /home/www/cvs/html/DIR
     	      $ ln -s /home/www/cvs/html/DIR RCS
     	      $ rcs-checkin .* *	# ignore warnings about . and ..
     	      $ rm RCS
     	      

     You are now done with this step. [@@This should be put into a script. -ttn] [@@And this script should walk the DIR doing this for each sub-dir from the bottom up. -len]

   • If you are adding a new subdirectory altogether, then check out a copy of the html module:

     	      $ [@@ Would a CVS Wizard PLEASE ... ]
     	      $ [@@ ... ADD SHELL COMMANDS etc. THAT DO THIS]
     	      

     create the directories and files you want, add a CVS exculsion file to not store certain files in CVS (for example, editor backup files, automatically generated files (e.g. TAGS) [@@ Would a CVS wizard please add information on how to do this -len] and then use

     	      $ cvs add
     	      

     to add them, working from the top to the leaves. Note that CVS can't handle symlinks; we'll put them it later. [@@This should be put into a script. -ttn] [@@And this script should walk the DIR doing this for each sub-dir from the bottom up. -len]

3. Once you have gotten the right portion of the CVS html module to look the way you want the corresponding portion of the web site to look, turn the appropriate portion of the web server's tree into a CVS working directory. Suppose SUBTREE has the form PARENT/TAIL, where TAIL is a single path component.

   [@@This should be put into a script. -ttn] [@@And this script should walk the DIR doing this for each sub-dir from the bottom up. -len]

   • Get a working dir, as a sibling of the directory you want to replace:

     	      $ cd /home/www/html/PARENT
     	      $ cvs checkout -d temp-dir html/SUBTREE
     	      

   • Move it into place:

     	      $ mv TAIL old-dir; mv temp-dir TAIL
     	      

   • CVS drops symlinks, so go back through your new tree and re-create any missing symlinks:

     	      find TAIL -type d ! -name CVS \
     	      | while read d; do
     	      ln -s `basename $d`.html $d/index.html
     	      done
     	      

     We'll just leave them there, and CVS will ignore them.

   • Use a web browser to check that the new tree looks right. Remember to reload the pages. If it's bogus, you still have your old, non-CVS copy.

4. Now that the web area is a CVS working tree, you need to tell CVS to update it whenever you commit a change. We do this by editing a file in the CVSROOT module that CVS uses to trigger actions based on the name of the directory and file being committed. Specifically, for files under html/SUBTREE, we want CVS to call the track-cvs script, which will update the web area CVS working tree. (NOTE: you need to be in Unix group cvsadmin to do this step.)

   [@@This should be put into a script. -ttn] [@@And this script should walk the DIR doing this for each sub-dir from the bottom up. -len]

   • Check out a working copy of the CVSROOT module:

     	      $ cd
     	      $ cvs checkout CVSROOT
     	      $ cd CVSROOT
     	      

   • Add the following line to the file loginfo in the current (CVSROOT) directory:

     	      ^html/SUBTREE $CVSROOT/CVSROOT/track-cvs /home/www/html/SUBTREE
     	      

   • Commit your changes to the file loginfo, so CVS can see them:

     	      $ cvs commit -m 'Automatically update SUBTREE' loginfo
     	      

5. Verify that the whole arrangement actually works:
   • Check out a new working copy of your subtree:

     	      $ cd
     	      $ cvs checkout html/SUBTREE
     	      

     Debugging Tip: In this step, make sure NOT to use

     	      $ cvs checkout -d DIR html/SUBTREE
     	      

     If you do so, CVS cannot match

     	      ^html/SUBTREE
     	      

     in the loginfo file and the live website will not be updated correctly.
     (This condition bit the guile pages and went uncorrected for three months! --ttn)

   • Commit some trivial change to some file in html/SUBTREE, and make sure that CVS correctly updated the web server's tree. For example:

     	      $ cd
     	      $ cvs edit html/SUBTREE/somefile.html
     	      $ echo >> html/SUBTREE/somefile.html
     	        # trivial change
     	        C-d
     	      $ cvs commit -m'testing' html/SUBTREE/somefile.html
     	      

     After some seconds, /home/www/cvs/html/SUBTREE/somefile.html should reflect the trivial change. You can check that file manually or with a web browser.

   • Compare the old and new directories with

     	      $ diff -r old-dir html/SUBTREE
     	      

     and fix any problems.

   • Once things are right, you can delete the old web tree:

     	      $ cd /home/www/html/PARENT
     	      $ rm -rf old-dir
     	      

   • At the end of each page that has this text, change

     	      Updated:
     	      <!-- hhmts start -->
     	      26 Jul 2000 tower
     	      <!-- hhmts end -->
     	      <HR>
     	      

     to

     	      Updated:
     	      <!-- timestamp start -->
     	      $Date$ $Author$
     	      <!-- timestamp end -->
     	      <HR>
     	      

     A way to do this, is to add the filename TAGS to the CVS exculsion file for html/SUBTREE. [@@ Would a CVS wizard please add information on how to do this -len] Then

     	      $ cd html/SUBTREE
     	      $ find . -name "*.html" -type f -print | xargs /usr/bin/etags -D  -l none -a -o TAGS
     	      

     Then use the GNU Emacs' function tags-query-replace to replace the text in each file that has it. Then

     	      $ rm html/SUBTREE/TAGS
     	      

Return to GNU's home page.

Please send FSF & GNU inquiries & questions to [email protected]. There are also other ways to contact the FSF.

Please send comments on these web pages to [email protected], send other questions to [email protected].

Copyright (C) 1999, 2000 Free Software Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110, USA

Verbatim copying and distribution of this entire article is permitted in any medium, provided this notice is preserved.

Updated: $Date: 2005/05/05 19:37:20 $ $Author: novalis $