Re: Backport non-gnulib dependent parts of my use-gnulib topic branch

2010-08-31 Thread Ralf Wildenhues
Hello Gary,

* Gary V. Vaughan wrote on Tue, Aug 31, 2010 at 08:43:14AM CEST:
 Okay to push?

For your information: I'm getting more reluctant to approve your
patches, based on how much I expect we have to fix regressions stemming
from them later on, when you are not present.

Please state if and how you tested your patches, each by itself or only
all in conjunction?  On what system(s)?  Did you use GNU tools in $PATH
or maybe ensure they are not found while testing?  Anything unusual in
your environment setup?  Do you try different versions of relevant tools
(autoconf, git)?  Whenever making things more complex rather than
simpler, do you have a rationale for doing so (besides that looked like
a cool thing to try out)?


Why do you post FYI commit emails without patches in them, and without
non-patch content over what gets auto-posted to the libtool-commit list?

Thanks,
Ralf



Re: Backport non-gnulib dependent parts of my use-gnulib topic branch

2010-08-31 Thread Gary V. Vaughan
Hallo Ralf,

On 1 Sep 2010, at 00:38, Ralf Wildenhues wrote:
 * Gary V. Vaughan wrote on Tue, Aug 31, 2010 at 08:43:14AM CEST:
 Okay to push?

Please ignore this patch series.  Cherry picking back to master from
a long divergent branch that has fairly intrusive reorganisation to
move Libtool to use gnulib doesn't work as well as I'd hoped.

I'll repost the entire series for consideration after the release.

 Please state if and how you tested your patches, each by itself or only
 all in conjunction?

On the original branch, each one had to pass `make distcheck' before
commit.

  On what system(s)?

For this series, on my development machine only, since the changees are
not machine specific.

 Did you use GNU tools in $PATH
 or maybe ensure they are not found while testing?

No.

 Anything unusual in your environment setup?

No.

 Do you try different versions of relevant tools (autoconf, git)?

git: no.
Autoconf, Automake: minimum supported versions and latest stable versions.

 Whenever making things more complex rather than
 simpler, do you have a rationale for doing so (besides that looked like
 a cool thing to try out)?

For this series, moving towards gnulib is the majority of the rationale,
and standardization of our use of scripts is the rest.

I assume you are asking why I rewrote the bootstrap script here - that was
because the existing script is out on a limb by itself, and I thought it
sensible to reuse the format and code from our other scripts.

 Why do you post FYI commit emails without patches in them, and without
 non-patch content over what gets auto-posted to the libtool-commit list?

Sorry, ./commit and ./libltdl/config/mailnotify run amok.  I haven't
really updated the latter to work properly with git.

Cheers,
-- 
Gary V. Vaughan (g...@gnu.org)


PGP.sig
Description: This is a digitally signed message part