Git for Squeeze
Hi!
I’m writing for advice. The old plan was to release with git 1.7.2.x,
but now that Squeeze is frozen I am having mild doubts.
1.7.2 has some regression fixes on top of the 1.7.1 that is in sid, in
addition to the countless tiny (lovely) usability enhancements.
The only risky bits from “git log --first-parent maint-1.7.1..maint”
I could find:
* gitweb: Use @diff_opts while using format-patch. The resulting
“rename” patches will not be applicable with old versions of GNU
patch.
That choice was made upstream, I think with good reason.
http://thread.gmane.org/gmane.comp.version-control.git/146822/focus=146841
For squeeze, until there is configuration available to override
this change, it might be worth leaving this out.
* jn/shortlog: respect --abbrev parameter in rev-list/log --format
callers. This caused some commands (e.g., “git commit”) to
temporarily stop abbreviating the commit IDs they print, but
I _think_ all problems of this kind have been tracked down and
fixed.
* by/log-follow: make log --follow check all files for predecessor
of a copy, not just the unmodified ones.
The patch creates a regression and a fix is ready on-list.
http://thread.gmane.org/gmane.comp.version-control.git/153470/focus=153488
* jk/url-decode: %-decode path part of ssh:// and file:// paths.
%-decoding is great but the +-decoding was not. There is a
patch in pu to fix that (730220de, Do not unquote + into ' '
in URLs, 2010-07-24).
* jn/gitweb-plackup: instaweb through plackup, static files in
separate directory.
This creates a static/ subdirectory in /usr/share/gitweb;
that would be great but I do not think now is the time to
iron out all the kinks in it.
I am considering preparing a package based on v1.7.2.1 that just
reverts those topics. Does that seem sane? Other preferences?
Regards,
Jonathan
Reply to: