[Monotone-devel] Re: cvsimport branch reconstruction status

[Bruce Stephens]

Markus Schiltknecht <address@hidden> writes:

> as you may know, I'm trying to implement the cvs2svn 2.0 algorithm
> inside monotones cvs_import command.

Which is cool: it would certainly be good to have a way to try and get
branches imported.

I confess that I think it might be better to do things in multiple
passes (with files inbetween) so that one can twiddle with the files
at suitable places.  CVS repositories tend to have messy bits which
might be better resolved by manual intervention.

However, I've no idea whether that's at all practical.  Quite probably
not.  The intermediate files that cvs2svn produces turned out to be
too big for me to fiddle with conveniently, and maybe that's
inevitable.

> I've made quite some progress recently, and just yesterday, my two
> test cases checking for correct branching succeeded.

It doesn't appear to compile (I tried
b9ad3efa71234bc4c83683ef63efb2fa765cefa7 and
3037a83cafbebaf21a8f32e19fca4c6e0b08b422).

[...]

> Another note: I've decided to use boost graph's topological sort
> before noting that monotone coded it's own graph stuff. Is it a bad
> idea to link to boost graph?

Yes, I suspect so.  Fewer dependencies on bits of boost the better, I
think.  On the other hand if boost graph provides significant
advantages then probably nobody will mind.