[Monotone-devel] Re: About the recent change of cvs_import...

[Bruce Stephens]

Nathaniel Smith <address@hidden> writes:

[...]

> Surely we can do better than this?  I _think_ cvs2svn does something
> more sensible here, and it's solving exactly the same problem we
> are; if someone could figure out how it handles this sort of
> wackiness, that would be very helpful.

The documentation is pretty reasonable, at least in outline.
<http://svn.collab.net/repos/cvs2svn/trunk/design-notes.txt>

I looked at how we might migrate from CVS (to what was, and is, still
in question) at work, and currently my plan is to use two or three of
the passes in cvs2svn, and probably hack the result by hand before
proceeding.  (Yes, it'll take forever to run, and the results will be
enormous, but so long as it's feasible to do the processing over a few
days we could do it.)

I think the problem is that in any sufficiently large CVS repository
you'll have things that can't possibly happen: invalid RCS files,
symbols that are both tags and branches, tag orders that are
inconsistent in different files, etc.  I suspect it's best to try and
do as much as possible automatically, but then to have some way to
punt significant decisions to humans, because even in CVS terms the
data's going to be impossible to interpret coherently.