Re: Emacs pretest 25.0.93

[phillip.lord@russet.org.uk (Phillip Lord)]

Eli Zaretskii <eliz@gnu.org> writes:

>> Date: Mon, 25 Apr 2016 21:25:59 +0800
>> From: Chris Zheng <chriszheng99@gmail.com>
>> Cc: "emacs-devel@gnu.org" <emacs-devel@gnu.org>
>> 
>> The attached is the screenshot running with `emacs -Q'.
>
> Unbelievable!
>
> AFAICT, the font used is actually bold-italic, not just italic, right?
> When you click S-mouse-1 and select "Change Buffer Font" from the menu
> that pops up, which font is shown as the current one?
>
> Anyway, I have no idea what could cause this, it certainly doesn't
> happen to me with the binary of the same tarball I built myself.  Does
> anyone else see that with 25.0.93 pretest (or the latest emacs-25
> branch)?  If not, I guess Phillip will have to see how this build was
> different from the previous one he did.

AFAICT, I did nothing different.

I had a go at bisecting this, and the results are so surprising, that I
need to replicate them on a different machine....


git bisect start
# bad: [5c587fdff164e8b90beb47f6da64b4884290e40a] Bump version to 25.0.93
git bisect bad 5c587fdff164e8b90beb47f6da64b4884290e40a
# good: [6287381da495ac5cc2a2a4acb7f49825677a44c8] ; Auto-commit of loaddefs files.
git bisect good 6287381da495ac5cc2a2a4acb7f49825677a44c8
# bad: [45577d548d3abfe504d3f936b3c9b997d0cf7f9c] Avoid GTK 3 crash with icons and masks
git bisect bad 45577d548d3abfe504d3f936b3c9b997d0cf7f9c
# good: [26b56dc739460bfd35bd12b4aca7434d381d9ef5] Fix some single quotes in documentation
git bisect good 26b56dc739460bfd35bd12b4aca7434d381d9ef5
# good: [9094304a9c109495bf2212a713df1b07270d15cd] * lisp/progmodes/xref.el (xref-buffer-name, xref--window): Move definitions before use.
git bisect good 9094304a9c109495bf2212a713df1b07270d15cd
# good: [d8b2ce5f8a6a849e55b48699dfa0e769a4b9a3f6] Ignore invalid base64 encoded embedded images
git bisect good d8b2ce5f8a6a849e55b48699dfa0e769a4b9a3f6
# good: [c0165ea4d6ecf81db6728782c7322c311ee0a783] Resurrect GNUS-NEWS autogeneration
git bisect good c0165ea4d6ecf81db6728782c7322c311ee0a783
# bad: [7eba90c12227d86636aaa3da8d68b0c213fb3910] Improve font selection by family on MS-Windows
git bisect bad 7eba90c12227d86636aaa3da8d68b0c213fb3910
# good: [9524ec5e4e3c7c3036f2640b2595d3755894c1b3] Ignore more merges when generating ChangeLog
git bisect good 9524ec5e4e3c7c3036f2640b2595d3755894c1b3
# bad: [73d213f2816876fe9c6c429e75a3be5454a42b34] Comint, term, and compile now set EMACS
git bisect bad 73d213f2816876fe9c6c429e75a3be5454a42b34
# first bad commit: [73d213f2816876fe9c6c429e75a3be5454a42b34] Comint, term, and compile now set EMACS


So, it's *that* commit again, setting EMACS=t. And here is the irony;
that commit has largely been reverted on current emacs-25.

I probably wont be able to test this on another machine today, so, I'm
sending this in case anyone has time to try the last two commits in the
bisect.

Phil