Re: Why I'm working on a NetBSD conversion

["Eric S. Raymond" <esr%thyrsus.com@localhost>]

S.P.Zeidler <spz%NetBSD.org@localhost>:
> A lot of work went into cleaning up syntax issues in the NetBSD CVS repo;
> due to its age, it contained a lot of traces of historical cvs bugs.

I have no doubt of it.  Give the size, I was expecting to see a lot more 
warning messages than I did.  There are about 1500 indications of possibly
dubious branch joins and a few other artifacts.

> The cvs currently used by NetBSD does changesets; I have no way of
> determining if these would be real enough for your tastes.

I'm aware that versions of GNU CVS since 1.12 issue commitids, of course.
But that only began in 2004 and most of your history will not have them.

> Also, if one really wanted to make the effort, we do have the commitlogs
> back to nearly the beginning of time, so the data could be extracted
> if one wanted to do that badly enough. That would also allow retracing
> subsequent repo moves.

That would be an *immense* amount of work, even with my tools.  Possibly
I could write some custom analyzers, but that is a degree of effort which
I would not undertake without having a very clear commitment that the
ewsults would be deployed.

> Regarding stress-testing targets for conversion, we have some files in
> the pkgsrc repo which stand out by the number of revisions.

Thank you, I will investigate this.

There remains the question of what NetBSD's strategic direction is.  Do you
intend to remain with CVS for the central repositories?  If not, can I
assist the transition out of it?
-- 
		<a href="http://www.catb.org/~esr/";>Eric S. Raymond</a>