* Overview

This file is a short instruction for maintainers on how to create and
publish the online version of the Tar Manual.

In the sections below we assume that the tar project has been properly
cloned from the git repo, bootstrapped and configured.  We also assume
that top-level directory of the project is the current local directory.

* Creating the web manual

To create the online version of the documentation, run

  make -C doc manual-rebuild

This will create the directory doc/manual populated with the tar
documentation files in various formats.  If the doc/manual directory
already exists, it will be removed prior to rebuilding.

The command produces very copious output.  We advise you to examine it
closely to make sure no error messages slip your attention. 

For the completeness sake, there are two more Makefile goals related
to the online manual:

** make -C doc clean-local

Removes the doc/manual directory, if it exists.

** make -C doc manual

Builds the doc/manual, unless it already exists.

* CVS Repository

The online tar manual[1] is a part of tar web pages[2] and is
traditionally maintained in the CVS repository[3].  To publish the
generated documentation, you will need first to check out tar web
pages from the CVS.  To do so, run

  cvs -z3 -d:ext:<username>@cvs.savannah.gnu.org:/web/tar co tar

where <username> is your user name on Savannah.  For the rest of this
document we will assume that the checked out version of the tar web
pages resides in the ~/websrc/tar directory.

If you have already checked out the web pages, be sure to update them
before publishing:

  cd ~/websrc/tar
  cvs update

* Publishing

To publish the created manual, change to the tar top-level directory
and run:

  rsync -avz --exclude CVS --delete manual ~/websrc/tar 

This will synchronize the newly created manual pages with the content
of the CVS sandbox.  Then, change to the ~/websrc/tar directory and
schedule any removed files for removal and any new files for addition
to the repository:

  cvs diff --brief 2>&1 | sed -n 's/.*cannot find //p' | xargs cvs rm
  cvs diff --brief 2>&1 | sed -n 's/^? //p' | xargs cvs add

Then commit your changes:

  cvs commit 

Once the changes are committed to CVS a job is scheduled on the server,
which synchronizes them with the content of the directory served by
the httpd daemon.  Normally such synchronization happens within
several seconds from the commit.

For more information about CVS, please see its documentation[4].

* References

[1] https://www.gnu.org/software/tar/manual/
[2] https://www.gnu.org/software/tar/
[3] https://web.cvs.savannah.gnu.org/viewvc/tar/
[4] https://www.nongnu.org/cvs/#documentation


Local Variables:
mode: outline
paragraph-separate: "[	]*$"
version-control: never
End: