[Monotone-devel] Re: A better approach for cvs->mtn?

[Bruce Stephens]

Markus Schiltknecht <address@hidden> writes:

[...]

> In monotone, you would use propagation to achieve the very
> same. Although, I'm unsure how cvs_import could handle that... maybe
> it's doable by manually telling it how to resolve *that* conflict.

No, propagate would also copy all the other changes from the trunk
(there's no restrictions version of propagate).  The closest would be
pluck, I guess, but that's equivalent to just making copies.

In reality I'm not sure we'd have done it at all in another system.
Or maybe we'd split the new code to a different branch, or something,
with appropriate twiddling of the release engineering scripts.  Or
just used pluck---probably we'd just have used pluck, I guess.  We did
it this way because CVS made it possible.

(Now I come to think of it, we do have suitable minimal tests we could
do: we have files containing information about the tags we used for
recent releases, so a test would be to try to reproduce all of those,
checking that the files only differ in RCS keywords.  They don't go
back that far, but maybe that would be enough (we used to use FLEXlm,
but now we don't, so we're unable to produce some releases any more
because we no longer have a license to use FLEXlm).)

[...]