Re: SIGSEGV on master when exporting MergedManuals.lyx in GUI
On Fri, Dec 03, 2021 at 11:04:23AM -0500, Scott Kostyshak wrote: > I have no idea what the root cause is. Reverting that commit, it seems > to show the output (note that this is regarding the default output, > which is HTML), but in the terminal I get the following, which I'm not > sure is relevant: > > ATTENTION: default value of option mesa_glthread overridden by environment. > ATTENTION: default value of option mesa_glthread overridden by environment. > ATTENTION: default value of option mesa_glthread overridden by environment. > ATTENTION: default value of option mesa_glthread overridden by environment. After reverting that commit, in addition to the above terminal messages, I also see: ###!!! [Parent][RunMessage] Error: Channel closing: too late to send/recv, messages will be lost Scott signature.asc Description: PGP signature -- lyx-devel mailing list lyx-devel@lists.lyx.org http://lists.lyx.org/mailman/listinfo/lyx-devel
Re: SIGSEGV on master when exporting MergedManuals.lyx in GUI
On Fri, Nov 05, 2021 at 12:10:20AM -0400, Scott Kostyshak wrote: > On Wed, Apr 14, 2021 at 12:45:04PM -0400, Scott Kostyshak wrote: > > On Tue, Apr 13, 2021 at 07:53:11PM +0100, José Abílio Matos wrote: > > > On Tuesday, April 13, 2021 5:17:24 PM WEST Scott Kostyshak wrote: > > > > I get the following terminal messages followed by a SIGSEGV when > > > > opening and > > > > viewing MergedManuals.lyx in the GUI: > > > > > > > > QSocketNotifier: Invalid socket 16 and type 'Read', disabling... > > > > > > > > Does anyone else see this on master? > > > > > > > > Scott > > > > > > For me it is socket 10. :-) > > > > > > LyX does not crash but it enters an infinite cycle (read that after 5 > > > minutes > > > it continues to output that line) sending that line to stderr. > > > I am not sure if I am patient enough to wait for infinity before > > > recovering > > > LyX's control. :-) > > > > Thanks for confirming, José. > > I can still reproduce the infinite cycle (but not the original SIGSEGV) on > master. Bisect leads to the following: 65b674ba4eff34df1ef7a7cf0f9e6a2585294c00 is the first bad commit commit 65b674ba4eff34df1ef7a7cf0f9e6a2585294c00 Author: Jean-Marc Lasgouttes Date: Thu Feb 18 15:35:42 2021 +0100 Fix a crash when closing tabs Although I do not know how to reproduce the crash, the change makes sense. This is backported from gadmm's lyx-unstable tree https://gitlab.com/gadmm/lyx-unstable/-/commit/261ce4ea98d82625a07f0fcf85e40f9055d2db3b I have no idea what the root cause is. Reverting that commit, it seems to show the output (note that this is regarding the default output, which is HTML), but in the terminal I get the following, which I'm not sure is relevant: ATTENTION: default value of option mesa_glthread overridden by environment. ATTENTION: default value of option mesa_glthread overridden by environment. ATTENTION: default value of option mesa_glthread overridden by environment. ATTENTION: default value of option mesa_glthread overridden by environment. I don't have a minimal document, but if you just edit MergedManuals.lyx to remove everything after the introduction.lyx child that reproduces the issue. Scott signature.asc Description: PGP signature -- lyx-devel mailing list lyx-devel@lists.lyx.org http://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Footnote size is larger on master (\normalsize is prepended)
On Fri, Dec 03, 2021 at 12:46:13PM +0100, Jean-Marc Lasgouttes wrote: > Le 31/10/2021 � 18:11, Scott Kostyshak a �crit�: > > See the attached file. On master, \normalsize is included inside the > > footnote, which is different from 2.3.x. > > Here is my take on how to properly fix this issue. I do not know however > what it does in other situations. Scott, do you have magic scripts and tests > to see the effect of this? No magic scripts. All I do is manually compile the 2.3.0 user documents with 2.3.0 (or 2.3.x) and with master and then run 'diffpdf' on the results. I'm not sure it makes sense to test that the LaTeX is equivalent since often there are differences in LyX's LaTeX output that are expected that do not cause a difference in the PDF output. We could imagine some automated tests with e.g. 'comparepdf' (a commandline tool) that I think could check whether the PDF output are the same. We'd have to disable the date showing up. Scott > > My belief is that this is the correct way of doing things. Some other parts > of the code may conflict with that, though. > > I added in the commit message some notes gathered during my research, so > that they do not get lost. > > JMarc > > -- > lyx-devel mailing list > lyx-devel@lists.lyx.org > http://lists.lyx.org/mailman/listinfo/lyx-devel signature.asc Description: PGP signature -- lyx-devel mailing list lyx-devel@lists.lyx.org http://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Footnote size is larger on master (\normalsize is prepended)
Le 31/10/2021 à 18:11, Scott Kostyshak a écrit :
See the attached file. On master, \normalsize is included inside the footnote,
which is different from 2.3.x.
Here is my take on how to properly fix this issue. I do not know however
what it does in other situations. Scott, do you have magic scripts and
tests to see the effect of this?
My belief is that this is the correct way of doing things. Some other
parts of the code may conflict with that, though.
I added in the commit message some notes gathered during my research, so
that they do not get lost.
JMarc
>From 83bfbaf2bd4a0672ee1222f5454859a2dba9db6b Mon Sep 17 00:00:00 2001
From: Jean-Marc Lasgouttes
Date: Fri, 3 Dec 2021 12:16:40 +0100
Subject: [PATCH] Tentative patch: fix font inside footnote inset
An inset that resets its font (like Footnote) does not care at all
about enclosing font. Therefore the real starting point is the class
default font. This avoid cases where the footnote contents is forced
to \normalsize.
Note that actually \normalfont resets everything but the font size.
This does not matter for footnote (which has its own font size), but
may matter elsewhere.
Concerning local_font: a research that excludes test and assignment
[*] shows that this is only used to remember language, which is a
different story (and not changed by this patch). The only exception
being in InsetMathHull::getCtObject and InsetMathNest::latex to
support change tracking in insets, but I am not 100% sure that this is
required. And historically [**] local_font used to be local_lang; it
may be good to return to this simpler variable later.
[*] git grep local_font src|grep -v 'local_font [!=]*='
[**] before afed118a, which improved support for InsetLine; however,
since 0e7ab29f, InsetLine does not depend on the current font anymore.
---
src/Paragraph.cpp | 7 +++
1 file changed, 3 insertions(+), 4 deletions(-)
diff --git a/src/Paragraph.cpp b/src/Paragraph.cpp
index 7fb46aee7d..f78925fafb 100644
--- a/src/Paragraph.cpp
+++ b/src/Paragraph.cpp
@@ -2533,10 +2533,9 @@ void Paragraph::latex(BufferParams const & bparams,
pos_type body_pos = beginOfBody();
unsigned int column = 0;
- // If we are inside an non inheritFont() inset, the real outerfont is local_font
- Font const real_outerfont = (!inInset().inheritFont()
- && runparams.local_font != nullptr)
- ? Font(runparams.local_font->fontInfo()) : outerfont;
+ // If we are inside an non inheritFont() inset, the outerfont is the default font
+ Font const real_outerfont =
+ inInset().inheritFont() ? outerfont : Font(bparams.documentClass().defaultfont());
if (body_pos > 0) {
// the optional argument is kept in curly brackets in
--
2.25.1
--
lyx-devel mailing list
lyx-devel@lists.lyx.org
http://lists.lyx.org/mailman/listinfo/lyx-devel
