gdb.git: author username-to-Full_Name+email map: RFC

[Jim Meyering <jim at meyering dot net>]

[resending, since the original was misclassified as spam by a bunch
 of spamassassin "chickenpox" rules.  I suspect it objected
 to the list of email addresses, so rather than including them
 inline, I'm using an attachment, now. ]

Hi,

I'm preparing a read-only git mirror of gdb's CVS repository.
It will be automatically sync'd every ~30 minutes, just as the other
http://sourceware.org/git/gitweb.cgi mirrors are: glibc, lvm2, dm,
automake[*].  This gdb.git mirror covers all branches and includes
all history.  For the record, I used a patched version of parsecvs to
do the initial conversion.  Automated incremental updates are handled
by a wrapper around "git cvsimport".

[*] Note that the automake.git mirror is obsolete, and will
be removed in a couple months.  It switched to git for good,
and is now hosted at savannah.org.

Part of the conversion job involves compiling the mapping from login name
(used in the cvs repository) to the "First Last <email>" that git uses.

Below, I've included a first cut at that mapping, produced mostly
automatically, partly using RealName+email from ChangeLog entries
committed by "login name".  The automatic process uses heuristics,
and is far from perfect.  I'm posting the result here, so people have
a chance to make corrections now.  It's important to get it right early,
because redoing the mapping requires changing nearly every SHA1 in the
resulting git repository, and that is very disruptive.

Note that this records the cvs committer name as the git "Author".
That is inaccurate when the cvs committer merely checked in a change
written by someone else.  If there is sufficient demand, we can look
into correcting those so that git shows the real "Author:" of the
proxy-committed changes.  That is possible only when the ChangeLog entry
includes the real author's name and email in an easily extracted form.

So if you can take a minute or two to sanity-check this
list, it may avoid some hassle latter:

Attachment: addr
Description: Binary data

Finally, note that with this list the only nameless
username is "gdbadmin".  However, any time a new
username cvs-commits a change, that username
will appear in the git repository with no associated
'First Last <email>' information.  If/when that happens,
let me know and I can update the map so that future
cvs commits include the usual information.
-----------------

Since I wrote the above, I've made the repository available:

Here's the gitweb interface:
  http://sourceware.org/git/gitweb.cgi?p=gdb.git

As it suggests at the top, you can get a copy via any
of these URLs:

  git://sources.redhat.com/git/gdb.git
  ssh://sources.redhat.com/git/gdb.git
  http://sources.redhat.com/git/gdb.git

E.g., run this

  git clone git://sources.redhat.com/git/gdb.git

That downloads ~73MB to create a copy of the repository in gdb/.git,
and puts all of the checked out files in gdb consuming a total of ~200MB.

If there are enough changes to the Authors list, I may regenerate
everything within the next couple weeks.

If any of you would like to rely on this, please ensure that all
of the files you care about are in the git repository.  For example,
I heard that three tcl-related directories need not be included, so I
excluded those along with the likes of ld, gold, gas, etc.

Also, please ensure that any branches you're interested in are
accurately reproduced in git.

Jim