Re: Converted git repository available for review

[Eric S. Raymond]

Paul Eggert <address@hidden>:
> Thanks.  I compared the latest revision (d245cfa) in the converted
> git repository's master to the corresponding revision (116808) in
> the bzr trunk, and found some discrepancies (see attached).  They
> fall into two major categories.
> 
> 1.  The git version is missing some changes to .gitignore files.
> For example, we removed the arch-tag lines a while ago, but the git
> version has resurrected them.  I expect the conversion machinery
> needs to be tweaked to handle .gitignore files correctly.

Take a look at the relevant section of the lift script.  What I've
done is treat the Bazaar ignore files as authoritative for those
revisions during which Bazaar was in use, ignoring .gitignores during
that period.  The other major possibility would be to simply remove
.bzrignores where .gitignore files exist.

I don't care.  Either is doable; somebody make a policy choice, please?

> 2.  The git version has rewritten 100 ChangeLog lines, replacing
> strings like "revno 108687" with strings like
> "2012-06-22T21:17:address@hidden" using the principles you
> discussed earlier.  In many cases the ChangeLog lines have gotten
> too long.  I propose converting them to the new form and fixing
> resulting line-length problems in the bzr trunk, now, so that the
> automated conversion process has nothing to do in this area.  I'll
> volunteer to do this by hand, as it hardly seems worth automating.

Here's I could do instead: select commit comments and ChangeLog
revisions with overlong lines, mailboxize them, and craft some filter
commands to fix up the line breaks.  That would fix up the history
for maximumm readability.

But how much do we care about fewer than 108 overlong lines, really?
I'm not even sure your small effort would be really justified, let
alone my larger one.
-- 
                <a href="http://www.catb.org/~esr/";>Eric S. Raymond</a>