[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [MIT-Scheme-devel] git?

From: Matt Birkholz
Subject: Re: [MIT-Scheme-devel] git?
Date: Sat, 2 May 2009 00:55:37 -0700

> From: address@hidden (Matt Birkholz)
> Date: Wed, 29 Apr 2009 17:12:12 -0700
> [...]
> History is dust.

And I am a history buff, apparently.  For whatever reason, I cannot
leave it alone.  I have found that this

    Argument "6521 FNK_SHOW_ALL" isn't numeric ...

is caused by my choice of cut-off tag, RELEASE_7_3_0, which is
"funky".  I can hack git-cvsimport to ignore the FNK_SHOW_ALL flag
(i.e. NOT skip the patchset), but the whole cut-off strategy is
leaving several files completely out of the dump -- files that have
not changed since the cut-off!

So... I looked again at fixing the broken files: hash.scm,v and
uenvir.scm,v.  I eventually got them patched so that all revisions
could be extracted AND they all have balanced parens.  With these
patches (attached), I can git-cvsimport ALL of src/runtime/, and the
resulting git HEAD is not missing any files.

The bit-rot in the repo. seems to start where there are changes near
the end of a revision that has no final newline.  Sound familiar?

I do not see any way (at the moment) to merge branches; the log
messages are not helpful.  I do not know where that leaves the unnamed
branches.  Without a named or merged tip, they are just left to
dangle, inaccessible?  Eventually they are GCed by git?

If losing the unnamed branches is acceptable, I would like to take a
crack at fixing the rest of the broken ,v files.  Can I get at the CVS
repo. myself somehow?  Can someone send me the rest of it?

Attachment: patch.diff.bz2
Description: CVS repo. fixes

reply via email to

[Prev in Thread] Current Thread [Next in Thread]