Call for testing: cvs2bzr

[Stephen J. Turnbull]

Greg Ward writes:

 > Reading fastimport from stdin sounds nice, but IMHO it's too damn
 > hard to be worth the effort.  git can presumably get away with it
 > because fastimport was designed to fit git's storage format.  bzr
 > and hg do not have that luxury, so our fastimporters have to do
 > ... something else.  And storing all blobs in memory is just not
 > scalable.

Well, if this works best for git, why not do cvs2git, then git2X?

Define a metadata format that fits into a git tag object (up to 8KiB,
IIRC), and have some pass emit a tag for *every* commit, containing
the metadata for X that you can deduce from CVS but git doesn't
include (is there any?)  (Of course you could also hack git itself to
include that metadata in commits.)  Then convert "tailor-style", by
repeatedly diffing and applying.  The special tags would have names
like cvs2bzr-$SHA1_OF_GIT_COMMIT so you could easily identify the
metadata for a given commit.

 > Anyways... getting off topic for cvs2svn list.  Sorry.

Scratched that addy.