Re: Hacks so I can daily-drive breakinsets3?
On Thu, Jul 09, 2026 at 12:27:18AM +0200, Jean-Marc Lasgouttes via lyx-devel wrote: > Le 08/07/2026 à 00:19, Jean-Marc Lasgouttes via lyx-devel a écrit : > > Le 08/07/2026 à 00:05, Scott Kostyshak a écrit : > > > > Do you have gdb set up so that you can give me a backtrace? > > > > > > Attached is the bt. > > > > Thanks. I have been reading the recent commits and identified this one > > as a possibility. > > This should be fixed, along with another potential crash. Thanks, I confirm I can no longer reproduce. Scott signature.asc Description: PGP signature -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
Le 08/07/2026 à 00:19, Jean-Marc Lasgouttes via lyx-devel a écrit : Le 08/07/2026 à 00:05, Scott Kostyshak a écrit : Do you have gdb set up so that you can give me a backtrace? Attached is the bt. Thanks. I have been reading the recent commits and identified this one as a possibility. This should be fixed, along with another potential crash. JMarc -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
Le 08/07/2026 à 00:05, Scott Kostyshak a écrit : Do you have gdb set up so that you can give me a backtrace? Attached is the bt. Thanks. I have been reading the recent commits and identified this one as a possibility. JMarc -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
On Tue, Jul 07, 2026 at 02:08:18PM +0200, Jean-Marc Lasgouttes via lyx-devel wrote: > Le 07/07/2026 à 14:06, Scott Kostyshak a écrit : > > On Tue, Jul 07, 2026 at 12:13:44AM +0200, Jean-Marc Lasgouttes via > > lyx-devel wrote: > > > Le 06/07/2026 à 22:35, Scott Kostyshak a écrit : > > > > > > 3. In the Beamer example document, press select all. I get the same > > > > issue I mentioned before. > > > > > > What issue? A crash? I do not see that. > > > > A crash. I just checked on current breakinsets3 (202fa449) and with a > > clean build, and I still get it: > > > > /usr/lib/gcc/x86_64-linux-gnu/15/../../../../include/c++/15/bits/stl_vector.h:1382: > > const_reference std::vector::back() const [_Tp = > > lyx::Row::Element, _Alloc = std::allocator]: Assertion > > '!this->empty()' failed. > > Aborted (core dumped) > > > > Perhaps to reproduce you would need to compile with -D_GLIBCXX_ASSERTIONS I > > think? > > Do you have gdb set up so that you can give me a backtrace? Attached is the bt. Scott (gdb) bt #0 __pthread_kill_implementation (threadid=, signo=6, no_tid=0) at ./nptl/pthread_kill.c:44 #1 __pthread_kill_internal (threadid=, signo=6) at ./nptl/pthread_kill.c:89 #2 __GI___pthread_kill (threadid=, signo=signo@entry=6) at ./nptl/pthread_kill.c:100 #3 0x75c4579e in __GI_raise (sig=sig@entry=6) at ../sysdeps/posix/raise.c:26 #4 0x75c288cd in __GI_abort () at ./stdlib/abort.c:73 #5 0x760ad3da in std::__glibcxx_assert_fail(char const*, int, char const*, char const*) () at /lib/x86_64-linux-gnu/libstdc++.so.6 #6 0x559adf86 in std::vector >::back (this=0x5703b510) at /usr/lib/gcc/x86_64-linux-gnu/15/../../../../include/c++/15/bits/stl_vector.h:1382 #7 0x559add39 in lyx::Row::back (this=0x5703b4e0) at /home/scott/lyxbuilds/breakinsets3/repo/src/Row.h:376 #8 0x559a9f82 in lyx::Row::isMarginSelected (this=0x5703b4e0, margin_begin=false, beg=..., end=...) at /home/scott/lyxbuilds/breakinsets3/repo/src/Row.cpp:293 #9 0x559aa280 in lyx::Row::setSelectionAndMargins (this=0x5703b4e0, beg=..., end=...) at /home/scott/lyxbuilds/breakinsets3/repo/src/Row.cpp:339 #10 0x55a32c33 in lyx::TextMetrics::setSelectionAndMargins (this=0x7fffe4002ec8, pit=5) at /home/scott/lyxbuilds/breakinsets3/repo/src/TextMetrics.cpp:2287 #11 0x55a26b01 in lyx::TextMetrics::drawParagraph (this=0x7fffe4002ec8, pi=..., pit=5, x=0, y=584) at /home/scott/lyxbuilds/breakinsets3/repo/src/TextMetrics.cpp:2359 #12 0x55a327e4 in lyx::TextMetrics::draw (this=0x7fffe4002ec8, pi=..., x=0, y=584) at /home/scott/lyxbuilds/breakinsets3/repo/src/TextMetrics.cpp:2248 #13 0x557e7a69 in lyx::BufferView::draw (this=0x567ef4e0, pain=..., paint_caret=false) at /home/scott/lyxbuilds/breakinsets3/repo/src/BufferView.cpp:3950 #14 0x55e9aef5 in lyx::frontend::GuiWorkArea::paintEvent (this=0x572e5ea0, ev=0x7fffb8b0) at /home/scott/lyxbuilds/breakinsets3/repo/src/frontends/qt/GuiWorkArea.cpp:1400 #15 0x777f26c5 in QWidget::event(QEvent*) () at /lib/x86_64-linux-gnu/libQt6Widgets.so.6 #16 0x7789cfca in QFrame::event(QEvent*) () at /lib/x86_64-linux-gnu/libQt6Widgets.so.6 #17 0x7657e0e8 in QCoreApplicationPrivate::sendThroughObjectEventFilters(QObject*, QEvent*) () at /lib/x86_64-linux-gnu/libQt6Core.so.6 #18 0x777980b8 in QApplicationPrivate::notify_helper(QObject*, QEvent*) () at /lib/x86_64-linux-gnu/libQt6Widgets.so.6 #19 0x55e09241 in lyx::frontend::GuiApplication::notify (this=0x566bb220, receiver=0x572e5d50, event=0x7fffb8b0) at /home/scott/lyxbuilds/breakinsets3/repo/src/frontends/qt/GuiApplication.cpp:3030 #20 0x765813f0 in QCoreApplication::notifyInternal2(QObject*, QEvent*) () at /lib/x86_64-linux-gnu/libQt6Core.so.6 #21 0x777e1d04 in QWidgetPrivate::sendPaintEvent(QRegion const&) () at /lib/x86_64-linux-gnu/libQt6Widgets.so.6 #22 0x777e438d in QWidgetPrivate::drawWidget(QPaintDevice*, QRegion const&, QPoint const&, QFlags, QPainter*, QWidgetRepaintManager*) () at /lib/x86_64-linux-gnu/libQt6Widgets.so.6 #23 0x7780669d in QWidgetRepaintManager::paintAndFlush() () at /lib/x86_64-linux-gnu/libQt6Widgets.so.6 #24 0x777f214a in QWidget::event(QEvent*) () at /lib/x86_64-linux-gnu/libQt6Widgets.so.6 #25 0x55e6003f in lyx::frontend::GuiView::event (this=0x56d921f0, e=0x56d8c2d0) at /home/scott/lyxbuilds/breakinsets3/repo/src/frontends/qt/GuiView.cpp:1978 #26 0x777980c8 in QApplicationPrivate::notify_helper(QObject*, QEvent*) () at /lib/x86_64-linux-gnu/libQt6Widgets.so.6 #27 0x55e09241 in lyx::frontend::GuiApplication::notify (this=0x566bb220, receiver=0x56d921f0, event=0x56d8c2d0) at /home/scott/lyxbuilds/breakinsets3/repo/src/frontends/qt/GuiApplication.cpp:3030 #28 0x765813f0 in QCoreApplication::notifyInternal2(QObject*, Q
Re: Hacks so I can daily-drive breakinsets3?
Le 07/07/2026 à 14:06, Scott Kostyshak a écrit : On Tue, Jul 07, 2026 at 12:13:44AM +0200, Jean-Marc Lasgouttes via lyx-devel wrote: Le 06/07/2026 à 22:35, Scott Kostyshak a écrit : 3. In the Beamer example document, press select all. I get the same issue I mentioned before. What issue? A crash? I do not see that. A crash. I just checked on current breakinsets3 (202fa449) and with a clean build, and I still get it: /usr/lib/gcc/x86_64-linux-gnu/15/../../../../include/c++/15/bits/stl_vector.h:1382: const_reference std::vector::back() const [_Tp = lyx::Row::Element, _Alloc = std::allocator]: Assertion '!this->empty()' failed. Aborted (core dumped) Perhaps to reproduce you would need to compile with -D_GLIBCXX_ASSERTIONS I think? Do you have gdb set up so that you can give me a backtrace? JMarc -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
On Tue, Jul 07, 2026 at 12:13:44AM +0200, Jean-Marc Lasgouttes via lyx-devel wrote: > Le 06/07/2026 à 22:35, Scott Kostyshak a écrit : > > 3. In the Beamer example document, press select all. I get the same > > issue I mentioned before. > > What issue? A crash? I do not see that. A crash. I just checked on current breakinsets3 (202fa449) and with a clean build, and I still get it: /usr/lib/gcc/x86_64-linux-gnu/15/../../../../include/c++/15/bits/stl_vector.h:1382: const_reference std::vector::back() const [_Tp = lyx::Row::Element, _Alloc = std::allocator]: Assertion '!this->empty()' failed. Aborted (core dumped) Perhaps to reproduce you would need to compile with -D_GLIBCXX_ASSERTIONS I think? Scott signature.asc Description: PGP signature -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
Le 06/07/2026 à 22:35, Scott Kostyshak a écrit : Thanks. The cursor placement now seems to work for me. Except there are still some issues: 1. In the example document I sent, put the cursor at the beginning of the buffer. Hold and press . Nothing happens. I see this. 2. In the example document I sent, anywhere in the frame title inset, hold . It seems to cycle back up to the top. This too. 3. In the Beamer example document, press select all. I get the same issue I mentioned before. What issue? A crash? I do not see that. JMarc -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
On Mon, Jul 06, 2026 at 05:29:49PM +0200, Jean-Marc Lasgouttes via lyx-devel wrote: > Le 06/07/2026 à 16:50, Jean-Marc Lasgouttes via lyx-devel a écrit : > > Note that I just pushed small changes that fix row alignment. > > These changes were what caused the crash :( > > Fixed now. Thanks. The cursor placement now seems to work for me. Except there are still some issues: 1. In the example document I sent, put the cursor at the beginning of the buffer. Hold and press . Nothing happens. 2. In the example document I sent, anywhere in the frame title inset, hold . It seems to cycle back up to the top. 3. In the Beamer example document, press select all. I get the same issue I mentioned before. Scott signature.asc Description: PGP signature -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
Le 06/07/2026 à 16:50, Jean-Marc Lasgouttes via lyx-devel a écrit : Note that I just pushed small changes that fix row alignment. These changes were what caused the crash :( Fixed now. JMarc -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
Le 06/07/2026 à 16:46, Scott Kostyshak a écrit : On Mon, Jul 06, 2026 at 04:35:37PM +0200, Jean-Marc Lasgouttes via lyx-devel wrote: Le 06/07/2026 à 15:54, Scott Kostyshak a écrit : Another issue I just found. To reproduce: 1. Open LyX. 2. Maximize the window. 3. Do select all. Result: I get the following: /usr/lib/gcc/x86_64-linux-gnu/15/../../../../include/c++/15/bits/stl_vector.h:1382: const_reference std::vector::back() const [_Tp = lyx::Row::Element, _Alloc = std::allocator]: Assertion '!this->empty()' failed. Can you reproduce? What document do you use? Sorry forgot to mention. This is the Beamer manual that is shipped with LyX: lib/examples/Presentations/Beamer.lyx OK. Note that for me, just loading your mwe crashes. Currently recompiling. Note that I just pushed small changes that fix row alignment. JMarc -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
On Mon, Jul 06, 2026 at 04:35:37PM +0200, Jean-Marc Lasgouttes via lyx-devel wrote: > Le 06/07/2026 à 15:54, Scott Kostyshak a écrit : > > Another issue I just found. To reproduce: > > > > 1. Open LyX. > > 2. Maximize the window. > > 3. Do select all. > > > > Result: I get the following: > > > > /usr/lib/gcc/x86_64-linux-gnu/15/../../../../include/c++/15/bits/stl_vector.h:1382: > > const_reference std::vector::back() const [_Tp = > > lyx::Row::Element, _Alloc = std::allocator]: Assertion > > '!this->empty()' failed. > > > > Can you reproduce? > > What document do you use? Sorry forgot to mention. This is the Beamer manual that is shipped with LyX: lib/examples/Presentations/Beamer.lyx Scott signature.asc Description: PGP signature -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
Le 06/07/2026 à 15:54, Scott Kostyshak a écrit : Another issue I just found. To reproduce: 1. Open LyX. 2. Maximize the window. 3. Do select all. Result: I get the following: /usr/lib/gcc/x86_64-linux-gnu/15/../../../../include/c++/15/bits/stl_vector.h:1382: const_reference std::vector::back() const [_Tp = lyx::Row::Element, _Alloc = std::allocator]: Assertion '!this->empty()' failed. Can you reproduce? What document do you use? JMarc -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
On Fri, Jul 03, 2026 at 09:00:23PM +0200, Jean-Marc Lasgouttes via lyx-devel wrote: > Le 11/05/2026 à 09:43, Scott Kostyshak a écrit : > > This looks great!! Thank you! > > > > A couple of issues: > > > > In Help > Beamer Presentations: > > > > if you hold the arrow, the cursor eventually gets stuck. > > Specifically, it gets stuck inside the title inset "Default overlay/action > > specifications vs" > > If I press "", it does not got down. > > I have been fixing stuff in the breakinsets3 branch these days, so that the > remaining issues are > - display of labels of floating insets is very basic. > - cursor positioning is incorrect in bidi documents > - input methods display may be broken (I trust Koji can fix that) > > I think it is time for a new round of testing. Another issue I just found. To reproduce: 1. Open LyX. 2. Maximize the window. 3. Do select all. Result: I get the following: /usr/lib/gcc/x86_64-linux-gnu/15/../../../../include/c++/15/bits/stl_vector.h:1382: const_reference std::vector::back() const [_Tp = lyx::Row::Element, _Alloc = std::allocator]: Assertion '!this->empty()' failed. Can you reproduce? Scott signature.asc Description: PGP signature -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
On Sat, Jul 04, 2026 at 09:30:55PM +0200, Jean-Marc Lasgouttes via lyx-devel wrote: > Le 04/07/2026 à 12:47, Scott Kostyshak a écrit : > > I attach a screenshot that shows a selection issue. Notice the cursor is > > not aligned with the selection. Since it seems you've already worked on > > selection (2ccda608), and I don't see it in your list of known issues, I > > report it. > > The first thing that I see is that the cursor is not where it should be. Can > you reproduce this without selecting? Yes, you are right. It's about cursor placement, not selection. > Other than that, the selection looks almost OK, except that the left margin > on the second row should be painted as selection. > > I'd appreciate it if you could provide a small self-contained file for me to > play with. I attach an MWE. Alternatively, to reproduce you can open up the Beamer example file, and just add random text to the frame title until you get 3 visual lines. Then the cursor issue shows up on the 2nd visual line. Can you reproduce now? Scott mwe-for-jmarc.lyx Description: application/lyx signature.asc Description: PGP signature -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
Am Samstag, dem 04.07.2026 um 21:52 +0200 schrieb Jean-Marc Lasgouttes via lyx-devel: > The problem is that there has to be enough room on the line for the > label. But one does not know where to break the line before knowing > which line should have the label on it. That makes things more > difficult. But if you say: * the longest INSETROW gets the label, and if there are multiple with the same length, the first of those this would almost always produce what I have proposed. And I would assume it is not very different from what we have now with non-flowing conglomerate insets, is it? > So I anticipate that having the label on the first line is easier > because it is deterministic. But I may be wrong and there may be > properties of the line breaking algorithm that makes life easier. You definitely know better, but I would think it could get tricky if the first line is very short. > Then I have the `small' issue that an inset in my model is > - an INSETBEGIN row element (the little corners on the left) > - one or more (when the inset is broken) INSETROW elements that point > to a row of text > - an INSETEND row element (little corners on the left) > > The spacing has to be added to the INSETBEGIN/END element and > actually depends on the INSETROW element (is the text there smaller > than the label string?). Therefore, getting the spacing right may be > tricky. > But I can make it work eventually. Do you know the length of each INSETROW element at this point? If so, it does not sound too hard (says the person who doesn't have to code it ;-) > > I am not sure whether what I write makes sense to you, but it helps > me to write down these ideas. Absolutely. -- Jürgen -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
Le 04/07/2026 à 13:22, Jürgen Spitzmüller a écrit : Looks really good! Thanks! * some extra whitespace around the insets so that there is enough room for the label. Note that this is (1) difficult to do and (2) can be ugly is an inset has a small part on one row and a small part n another row and the label is a bit long. ... this would be preferable. This is what we currently have with "conglomerate" insets. If the label is only given on the longest line (see below), it doesn't sound more difficult than what we have now (but it might be, of course) The problem is that there has to be enough room on the line for the label. But one does not know where to break the line before knowing which line should have the label on it. That makes things more difficult. So I anticipate that having the label on the first line is easier because it is deterministic. But I may be wrong and there may be properties of the line breaking algorithm that makes life easier. Then I have the `small' issue that an inset in my model is - an INSETBEGIN row element (the little corners on the left) - one or more (when the inset is broken) INSETROW elements that point to a row of text - an INSETEND row element (little corners on the left) The spacing has to be added to the INSETBEGIN/END element and actually depends on the INSETROW element (is the text there smaller than the label string?). Therefore, getting the spacing right may be tricky. But I can make it work eventually. I am not sure whether what I write makes sense to you, but it helps me to write down these ideas. JMarc -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
Le 04/07/2026 à 12:47, Scott Kostyshak a écrit : I attach a screenshot that shows a selection issue. Notice the cursor is not aligned with the selection. Since it seems you've already worked on selection (2ccda608), and I don't see it in your list of known issues, I report it. The first thing that I see is that the cursor is not where it should be. Can you reproduce this without selecting? Other than that, the selection looks almost OK, except that the left margin on the second row should be painted as selection. I'd appreciate it if you could provide a small self-contained file for me to play with. JMarc -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
Am Freitag, dem 03.07.2026 um 21:00 +0200 schrieb Jean-Marc Lasgouttes via lyx-devel: > I have been fixing stuff in the breakinsets3 branch these days, so > that the remaining issues are > - display of labels of floating insets is very basic. > - cursor positioning is incorrect in bidi documents > - input methods display may be broken (I trust Koji can fix that) Looks really good! > I think it is time for a new round of testing. I need everybody's > thoughts on : > > - is it better to have > * a label that spills on neighbors Would be acceptable IMHO, however ... > * a label that is removed when there is not enough room > * a label that is cut when there is not enough room > * some extra whitespace around the insets so that there is enough > room for the label. Note that this is (1) difficult to do and (2) can > be ugly is an inset has a small part on one row and a small part n > another row and the label is a bit long. ... this would be preferable. This is what we currently have with "conglomerate" insets. If the label is only given on the longest line (see below), it doesn't sound more difficult than what we have now (but it might be, of course) > > - shall the label be > * on each line (like now) > * on the first line > * on the longest line This (the first longest line, for that matter) Thanks, -- Jürgen -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
On Fri, Jul 03, 2026 at 09:00:23PM +0200, Jean-Marc Lasgouttes via lyx-devel wrote: > I have been fixing stuff in the breakinsets3 branch these days, so that the > remaining issues are > - display of labels of floating insets is very basic. > - cursor positioning is incorrect in bidi documents > - input methods display may be broken (I trust Koji can fix that) > > I think it is time for a new round of testing. Great! Thank you. I found one issue (reported below). > I need everybody's thoughts > on : > > - is it better to have > * a label that spills on neighbors > * a label that is removed when there is not enough room > * a label that is cut when there is not enough room > * some extra whitespace around the insets so that there is enough room for > the label. Note that this is (1) difficult to do and (2) can be ugly is an > inset has a small part on one row and a small part n another row and the > label is a bit long. Not sure. > - shall the label be > * on each line (like now) > * on the first line > * on the longest line > > Note that if there is no label for a given part of flowing inset, it is > possible to have a tooltip that gives the information. I'll think about this. I attach a screenshot that shows a selection issue. Notice the cursor is not aligned with the selection. Since it seems you've already worked on selection (2ccda608), and I don't see it in your list of known issues, I report it. Scott signature.asc Description: PGP signature -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
Le 11/05/2026 à 09:43, Scott Kostyshak a écrit : This looks great!! Thank you! A couple of issues: In Help > Beamer Presentations: if you hold the arrow, the cursor eventually gets stuck. Specifically, it gets stuck inside the title inset "Default overlay/action specifications vs" If I press "", it does not got down. I have been fixing stuff in the breakinsets3 branch these days, so that the remaining issues are - display of labels of floating insets is very basic. - cursor positioning is incorrect in bidi documents - input methods display may be broken (I trust Koji can fix that) I think it is time for a new round of testing. I need everybody's thoughts on : - is it better to have * a label that spills on neighbors * a label that is removed when there is not enough room * a label that is cut when there is not enough room * some extra whitespace around the insets so that there is enough room for the label. Note that this is (1) difficult to do and (2) can be ugly is an inset has a small part on one row and a small part n another row and the label is a bit long. - shall the label be * on each line (like now) * on the first line * on the longest line Note that if there is no label for a given part of flowing inset, it is possible to have a tooltip that gives the information. JMarc -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
On Sun, May 10, 2026 at 08:00:08PM +0200, Jean-Marc Lasgouttes via lyx-devel wrote: > Le 10/05/2026 à 10:56, Scott Kostyshak a écrit : > > For example, when I open the math manual, if I search for "Here is an > > example" and put the cursor anywhere in there and then hold , > > after 10 seconds of holding it I get a SIGSEGV (I assume because of the > > math just below). > > I see it too, and it is because posVisRight is invoked after a > LFUN_FINISHED_RIGHT. I fixed this code in master and adopted the change in > brkinset3 (I rebase often anyway). Thanks! > > Only make that quick'n'dirty version if it wouldn't take too much time > > and you think it would be useful (to you) for me to test the other parts > > of the code. If not, tell me "fiche-moi la paix!" and I promise I will > > be patient :) > > The quick and dirty cure is in. Two main flaws: > 1/ the label is on every row (tell me whether this is a problem and where it > should be if there is only one: first, largest, middle...) > > 2/ the label will spill on neighbor. This will probably be the most annoying > to solve (although the other one needs thought too). This looks great!! Thank you! A couple of issues: In Help > Beamer Presentations: if you hold the arrow, the cursor eventually gets stuck. Specifically, it gets stuck inside the title inset "Default overlay/action specifications vs" If I press "", it does not got down. Perhaps related to the above, but if I keep opening Beamer manually and holding , (and 'ing out when I get stuck) eventually I get a SIGABRT. Sometimes this takes me a few minutes to trigger. I get the following backtrace: Thread 1 "lyx" received signal SIGABRT, Aborted. __pthread_kill_implementation (threadid=, signo=6, no_tid=0) at ./nptl/pthread_kill.c:44 warning: 44 ./nptl/pthread_kill.c: No such file or directory (gdb) bt #0 __pthread_kill_implementation (threadid=, signo=6, no_tid=0) at ./nptl/pthread_kill.c:44 #1 __pthread_kill_internal (threadid=, signo=6) at ./nptl/pthread_kill.c:89 #2 __GI___pthread_kill (threadid=, signo=signo@entry=6) at ./nptl/pthread_kill.c:100 #3 0x75c4579e in __GI_raise (sig=sig@entry=6) at ../sysdeps/posix/raise.c:26 #4 0x75c288cd in __GI_abort () at ./stdlib/abort.c:73 #5 0x5593773b in lyx::lyx_exit (exit_code=1) at /home/scott/lyxbuilds/master-master/repo/src/LyX.cpp:262 #6 0x56237242 in lyx::doAssertWithCallstack (value=false) at /home/scott/lyxbuilds/master-master/repo/src/support/lassert.cpp:50 #7 0x5623790e in lyx::doAssert (expr=0x5626ebf1 "hasPosition()", file=0x5626eb7d "/home/scott/lyxbuilds/master-master/repo/src/ParagraphMetrics.cpp", line=68) at /home/scott/lyxbuilds/master-master/repo/src/support/lassert.cpp:59 #8 0x559a158e in lyx::ParagraphMetrics::position (this=0x594938d8) at /home/scott/lyxbuilds/master-master/repo/src/ParagraphMetrics.cpp:68 #9 0x557e8e99 in lyx::ParagraphMetrics::top (this=0x594938d8) at /home/scott/lyxbuilds/master-master/repo/src/ParagraphMetrics.h:96 #10 0x55a2cff2 in lyx::TextMetrics::getPitNearY (this=0x59438868, y=1028, strict=false) at /home/scott/lyxbuilds/master-master/repo/src/TextMetrics.cpp:1809 #11 0x55a2d1b4 in lyx::TextMetrics::getRowNearY (this=0x59438868, y=1028, strict=false) at /home/scott/lyxbuilds/master-master/repo/src/TextMetrics.cpp:1846 #12 0x55a2d35e in lyx::TextMetrics::editXY (this=0x59438868, cur=..., x=909, y=1028) at /home/scott/lyxbuilds/master-master/repo/src/TextMetrics.cpp:1868 #13 0x55db3929 in lyx::InsetText::editXY (this=0x584183e0, cur=..., x=909, y=1028) at /home/scott/lyxbuilds/master-master/repo/src/insets/InsetText.cpp:317 #14 0x55c53aa5 in lyx::InsetCollapsible::editXY (this=0x584183e0, cur=..., x=909, y=1028) at /home/scott/lyxbuilds/master-master/repo/src/insets/InsetCollapsible.cpp:646 #15 0x55a2d4cb in lyx::TextMetrics::editXY (this=0x59412318, cur=..., x=909, y=1028) at /home/scott/lyxbuilds/master-master/repo/src/TextMetrics.cpp:1894 #16 0x55db3929 in lyx::InsetText::editXY (this=0x58417d30, cur=..., x=909, y=1028) at /home/scott/lyxbuilds/master-master/repo/src/insets/InsetText.cpp:317 #17 0x55c53aa5 in lyx::InsetCollapsible::editXY (this=0x58417d30, cur=..., x=909, y=1028) at /home/scott/lyxbuilds/master-master/repo/src/insets/InsetCollapsible.cpp:646 #18 0x55a2d4cb in lyx::TextMetrics::editXY (this=0x574c7938, cur=..., x=909, y=1028) at /home/scott/lyxbuilds/master-master/repo/src/TextMetrics.cpp:1894 #19 0x5584a730 in lyx::Cursor::upDownInText (this=0x58b69638, up=false) at /home/scott/lyxbuilds/master-master/repo/src/Cursor.cpp:2269 Scott signature.asc Description: PGP signature -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
Le 10/05/2026 à 10:56, Scott Kostyshak a écrit : For example, when I open the math manual, if I search for "Here is an example" and put the cursor anywhere in there and then hold , after 10 seconds of holding it I get a SIGSEGV (I assume because of the math just below). I see it too, and it is because posVisRight is invoked after a LFUN_FINISHED_RIGHT. I fixed this code in master and adopted the change in brkinset3 (I rebase often anyway). Only make that quick'n'dirty version if it wouldn't take too much time and you think it would be useful (to you) for me to test the other parts of the code. If not, tell me "fiche-moi la paix!" and I promise I will be patient :) The quick and dirty cure is in. Two main flaws: 1/ the label is on every row (tell me whether this is a problem and where it should be if there is only one: first, largest, middle...) 2/ the label will spill on neighbor. This will probably be the most annoying to solve (although the other one needs thought too). JMarc -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
On Sat, May 09, 2026 at 10:56:19PM +0200, Jean-Marc Lasgouttes via lyx-devel wrote: > Le 08/05/2026 à 11:06, Scott Kostyshak a écrit : > > I'm currently daily-driving master branch. > > > > Scrolling through my documents on breakinsets3, things look great! Looks > > quite smooth, and better than master branch thanks to the flowing insets > > feature. > > > > I'd like to daily-drive breakinsets3, but there are a couple of issues > > that make it hard, which I assume are known and difficult to fix > > correctly: > > > > 1. Cursor navigation (I forget whether with mouse or keyboard). > > Do you use logical cursor movement or visible cursor movement? The latter is > broken now, but makes no difference in LtR documents. Right-to-left cursor movement is set to "Logical", although indeed I do not use RtL. I think the issue is because of math. For example, when I open the math manual, if I search for "Here is an example" and put the cursor anywhere in there and then hold , after 10 seconds of holding it I get a SIGSEGV (I assume because of the math just below). > > 2. Labels. > > > > Is there any hack I can apply on my local branch that would make it so I > > can daily-drive? For example, for labels, would it be not-too-hard to at > > least paint the label for the special case in which there is no > > linebreak? If so, that might be enough for me to daily-drive it. > > > > No problem if there's nothing easy to do about this. I'm just itching to > > play around with the nice feature while I work :) > > The only difficulty is time right now. This is a very valid reason. > But I will make a quick'n'dirty > version for your entertainment. Only make that quick'n'dirty version if it wouldn't take too much time and you think it would be useful (to you) for me to test the other parts of the code. If not, tell me "fiche-moi la paix!" and I promise I will be patient :) Scott signature.asc Description: PGP signature -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
Re: Hacks so I can daily-drive breakinsets3?
Le 08/05/2026 à 11:06, Scott Kostyshak a écrit : I'm currently daily-driving master branch. Scrolling through my documents on breakinsets3, things look great! Looks quite smooth, and better than master branch thanks to the flowing insets feature. I'd like to daily-drive breakinsets3, but there are a couple of issues that make it hard, which I assume are known and difficult to fix correctly: 1. Cursor navigation (I forget whether with mouse or keyboard). Do you use logical cursor movement or visible cursor movement? The latter is broken now, but makes no difference in LtR documents. 2. Labels. Is there any hack I can apply on my local branch that would make it so I can daily-drive? For example, for labels, would it be not-too-hard to at least paint the label for the special case in which there is no linebreak? If so, that might be enough for me to daily-drive it. No problem if there's nothing easy to do about this. I'm just itching to play around with the nice feature while I work :) The only difficulty is time right now. But I will make a quick'n'dirty version for your entertainment. JMarc -- lyx-devel mailing list [email protected] https://lists.lyx.org/mailman/listinfo/lyx-devel
