New Emacs Bazaar Repository II

[Stephen J. Turnbull]

Ian Clatworthy writes:

 > Wow - that's weird/interesting. fast-import-info is typically really
 > fast. I tested it on the Linux kernel the other day and it only took
 > 17 minutes. (Which sounds long but isn't given the 300K blobs and 130K
 > revisions in that f-i stream).

It still sounds long.  Have you considered breaking out the structure
info (git's refs, commits, and trees) from the stream of blobs as
cvs2svn does (at least in cvs2git mode)?  That seems really close to
what Jason did, actually.

With that structure, "git fast-import" on the XEmacs fast-import
stream takes about 80 seconds, less than twice the time as "grep -c
blob git-blob.dat", which takes 55 seconds (and tells me I have 52114
blobs, "grep -c commit git-dump.dat" says 18784 revisions).  Scaling
up to the Linux kernel size, you ought to be able to halve that time
(or better on a decent file system: my tests are on a Mac).  I don't
think there's an equivalent command for fast-import-info, so I can't
test that, but working on the dump.dat data only should make it
blindingly fast, a handful of seconds, I should think.