Find replace history not saved between sessions

[DiapDealer]

No it doesn't help. I can't get any debug output out of the destructor OR the close() slot. Either the segfault is happening first (though I can't see any other missing data from Sigil.ini at first glance), or something else is going on.

It doesn't make any sense that I can't get any debug output written to stdout from the close() slot. Because the WriteSettingsVisible(false) IS being correctly written to the ini file from the close() slot. So I can't say for certain whether the destructor is being called or not.

[poxi1023]

Originally Posted by KevinH

What happens if you manually close the Find and Replace window after trying some searches before closing Sigil? After the manual close writesetting should tell SettingsStore to update itself. Please check the sigil.ini file before quitting out of Sigil to see if anything is being written.

I checked as you said, but nothing is written. find_strings and replace_strings remains always empty.

[poxi1023]

Originally Posted by DiapDealer

No it doesn't help. I can't get any debug output out of the destructor OR the close() slot. Either the segfault is happening first (though I can't see any other missing data from Sigil.ini at first glance), or something else is going on.

It doesn't make any sense that I can't get any debug output written to stdout from the close() slot. Because the WriteSettingsVisible(false) IS being correctly written to the ini file from the close() slot. So I can't say for certain whether the destructor is being called or not.

Now I started Sigil from the CLI. Did some searches, as always nothing written in sigil.ini. I got an output, which I don't understand, but maybe it means something to you guys.

[DiapDealer]

OK. Very weird. I don't think the segfault is connected.

It's the ToC QDockWidget for whatever reason. If it's visible, the settings won't get written to disk and the destructor never fires. You have to close the ToC dockwidget, close Sigil (still no settings written at this point), and then relaunch Sigil (with the toc dockwidget closed). When next closed, all events fire, all debug output prints to stdout, and the F&R settings all get flushed to disk. All will work normally --- until the Table of Contents dockwidget is made visible again. Then it's lather, rinse, repeat.

Curious to see if poxi1023 can confirm the same behavior.

[KevinH]

The last line of your output indicates a segfault is happening for you too.

Originally Posted by poxi1023

Now I started Sigil from the CLI. Did some searches, as always nothing written in sigil.ini. I got an output, which I don't understand, but maybe it means something to you guys.

[poxi1023]

Originally Posted by DiapDealer

OK. Very weird. I don't think the segfault is connected.

It's the ToC QDockWidget for whatever reason. If it's visible, the settings won't get written to disk and the destructor never fires. You have to close the ToC dockwidget, close Sigil (still no settings written at this point), and then relaunch Sigil (with the toc dockwidget closed). When next closed, all events fire, all debug output prints to stdout, and the F&R settings all get flushed to disk. All will work normally --- until the Table of Contents dockwidget is made visible again. Then it's lather, rinse, repeat.

Curious to see ifpoxi1023 can confirm the same behavior.

Yes, that is it! With the ToC window closed, Sigil behaves well.

[DiapDealer]

Originally Posted by poxi1023

Now I started Sigil from the CLI. Did some searches, as always nothing written in sigil.ini. I got an output, which I don't understand, but maybe it means something to you guys.

That's platform theme stuff, I think. From the CLI type "unset QT_QPA_PLATFORMTHEME" (without the quotes) before typing "sigil" and see if that output goes away. I always do that when debugging to make sure none of the platform extras are getting in the way. You're only unsetting the variable for that cli session, so there's no worries. Close it, open another terminal, and it will all be back to the way it was.

[KevinH]

Perhaps never having the TOC dockwidget open prevents the segfault allowing things to work?

[DiapDealer]

Originally Posted by poxi1023

Yes, that is it! With the ToC window closed, Sigil behaves well.

Thanks for confirming. I still get the segfault in the terminal windows whenever Sigil closes, so I'm assuming you do too?

[DiapDealer]

Originally Posted by KevinH

Perhaps never having the TOC dockwidget open prevents the segfault allowing things to work?

I won't swear to it just yet, but I think the the segfault is unrelated. I still get the segfault when settings are being properly written.

The gdb backtrace on the segfault shows it to be something in QtWebEngineWidgets. But without debug versions of Qt/QtWebEngine 5.9.5, I'm unlikely to get much more detail than that.

[poxi1023]

Originally Posted by DiapDealer

Thanks for confirming. I still get the segfault in the terminal windows whenever Sigil closes, so I'm assuming you do too?

I do too. New searches get remembered, as long as the ToC window remains closed. I have to say that I use that window quite regularly, so I'd prefer some other solution, if possible.

[DiapDealer]

Originally Posted by poxi1023

I have to say that I use that window quite regularly, so I'd prefer some other solution, if possible.

We'll do what we can, but it's clear that it's not a problem with later versions of Qt. So it may not be something that's within our control.

[KevinH]

FWIW, Qt 5.12.7 has now been released and has a number of bug fixes. Would an official Qt binary stand any chance of working with these Linux distributions?

[KevinH]

BTW, is the TOC dockwidget the only one that has impact? Does closing other dockwidgets help?

[DiapDealer]

@pazos (earlier in the thread) is building Sigil against Qt 5.14.x on the same version of Ubuntu as poxi1023. I'm assuming he's using a binary version from Qt.