2018-02-24 GnuCash IRC logs

00:17:07 *** wget has quit IRC
00:17:41 *** wget has joined #gnucash
00:56:42 *** O01eg has quit IRC
01:02:37 *** Mechtilde has joined #gnucash
01:08:18 *** fell has quit IRC
01:09:23 *** fell has joined #gnucash
01:13:14 *** pilotauto has joined #gnucash
01:13:55 *** darshan has joined #gnucash
01:14:56 *** Cuare has quit IRC
01:17:56 *** O01eg has joined #gnucash
01:44:42 *** darshan has quit IRC
01:53:59 *** darshan has joined #gnucash
01:58:19 *** darshan has quit IRC
02:08:16 *** gjanssens has joined #gnucash
02:08:16 *** ChanServ sets mode: +o gjanssens
02:19:21 <gjanssens> .
02:59:54 *** woodrec has joined #gnucash
03:38:24 <Mechtilde> GOOD MORGNIN
03:38:29 <Mechtilde> sorry
03:38:41 <Mechtilde> good morning
03:39:39 <Mechtilde> I correct my transactions and so on. I have to delete some double entries. most of the time GnuCash crashes
03:40:41 <Mechtilde> it's still 2.7.3
03:59:47 *** gncbot sets mode: +o fell
03:59:49 <fell> Moin
04:00:49 <fell> Crash while deleting txns in the register view, Mechtilde?
04:01:26 <Mechtilde> yes
04:03:33 <Mechtilde> the deletion itself is completed
04:04:01 <Mechtilde> and it crashes 8 of 10 times
04:04:51 <fell> Is it self compiled or from some package manager?
04:05:04 <Mechtilde> it is from Debian
04:05:30 <Mechtilde> yes from Debian experimental
04:06:02 <fell> So a previous version should be clean uninstalled, or do you have a stable version in parallel?
04:06:46 <Mechtilde> the previous version is deinstalled
04:06:59 <Mechtilde> no parallel version
04:09:02 <fell> Do you think you can install the respective -dbg packages and create a https://wiki.gnucash.org/wiki/Stack_Trace
04:10:36 <Mechtilde> there is no -dbg package for 2.7.3
04:10:47 <Mechtilde> I will ask the maintainer to provide one
04:12:36 <fell> It might also already been included on their experimental branch.
04:14:02 <Mechtilde> https://packages.debian.org/search?keywords=gnucash&searchon=names&suite=all§ion=all&sourceid=mozilla-search
04:14:25 <Mechtilde> thre is only nucash-dbgsym for alpha
04:14:39 <Mechtilde> +g
04:15:47 <Mechtilde> I will try it under gdb
04:17:02 <Mechtilde> if I will have some txns to be deleted
04:24:37 <fell> Did you watch something they had in common like multisplit, foreign currency, ...?
04:30:42 <Mechtilde> no multisplit
04:30:46 <Mechtilde> normal txns
04:40:10 <Mechtilde> I started a new session with gdb
04:40:52 <Mechtilde> it hangs with:
04:41:05 <Mechtilde> (gnucash:11306): Gtk-WARNING **: Negative content width -12 (allocation 1, extents 6x7) while allocating gadget (node button, owner GtkButton)
04:41:05 <Mechtilde> Thread 1 "gnucash" received signal SIGSEGV, Segmentation fault.
04:41:05 <Mechtilde> 0x00007fffef932bd6 in xaccTransCountSplits ()
04:41:05 <Mechtilde> from /usr/lib/x86_64-linux-gnu/gnucash/gnucash/libgncmod-engine.so
04:41:26 <Mechtilde> (gdb)
04:44:21 <Mechtilde> I will provide a bug report
04:44:55 <fell> Something like that is also in https://bugzilla.gnome.org/show_bug.cgi?id=793771
04:45:24 <fell> Perhaps gjanssens knows more?
04:46:22 *** pilotauto has quit IRC
04:47:06 <Mechtilde> I can provide my backlog too
04:47:57 <gjanssens> fell: unfortunately I don't know more. The bug you point at is worrying me because if it's right what the reporter says, gnucash can't currently run under wayland (the modern compositor all distros are slowly migrating to)
04:48:14 <gjanssens> Or more precisely online banking seems to have issues there.
04:48:30 <gjanssens> Mechtilde: yes, it would be interesting to see your full backtrace.
04:48:52 <Mechtilde> yes I did bt full
04:54:12 <Mechtilde> it is in the bugzilla
04:54:24 <Mechtilde> https://bugzilla.gnome.org/show_bug.cgi?id=793771
04:56:51 <Mechtilde> gjanssens, I forgot to say that I use XFCE4
04:58:35 <gjanssens> Thanks Mechtilde
05:01:18 *** tong has quit IRC
05:01:22 <Mechtilde> gjanssens, is it reasonable to do the same for the SEPA txns, which I show you in Brussels?
05:02:18 <gjanssens> Mechtilde: your backtrace is missing debug symbols, making it less useful. Can you reconfigure your gnucash build adding "-DCMAKE_BUILD_TYPE=debug" to your cmake command ?
05:02:37 <Mechtilde> it isn't my own build
05:02:47 <gjanssens> But I can already see it's not related ot bug 793771. It's best to open a new bug report with it.
05:02:54 <gjanssens> Ok.
05:02:58 <gjanssens> Where did you get it ?
05:03:10 <Mechtilde> and Debian didn't provide the -dgb package for version 2.7.3
05:03:21 <gjanssens> Hmm, that's unfortunate.
05:03:34 <Mechtilde> I can do an own bug report
05:04:01 <gjanssens> Can you request a dbg package ? For a development build that's even more useful than for stable builds.
05:04:05 <gjanssens> Ok thanks.
05:04:09 <Mechtilde> https://packages.debian.org/search?keywords=gnucash&searchon=names&suite=all§ion=all&sourceid=mozilla-search
05:05:13 <gjanssens> Mechtilde: there is package -dbgsym which seems to provide the debug symbols for gnucash
05:05:37 <Mechtilde> but for 2.73 it is only for alpha
05:05:45 <Mechtilde> not for amd64
05:07:12 <gjanssens> Oh, I missed that.
05:07:26 <gjanssens> Can you contact the debian packager about this ?
05:08:21 <Mechtilde> I try it
05:12:13 <gjanssens> Ok.
05:12:21 *** gjanssens is now known as gjanssens_afk
05:24:37 <Mechtilde> gjanssens_afk, I found https://bugzilla.gnome.org/show_bug.cgi?id=792763 too
05:25:29 <Mechtilde> this seems to be more similar to mine
05:43:50 <chris> gjanssens_afk jralls - I think one of my architectural decisions in TR-upgrade needs to be revised - when I was creating new multichoices e.g. filter/reconcile status, general/infobox, I thought I would be clever and store the value as a list. e.g. (list #\n #\c #\y) which can be queried directly. This was not standard behaviour.
05:44:20 <chris> From own UI testing it works well, and seems to be persistent across sessions / saved reports etc.
05:46:58 <chris> However I find it impossible to (gnc:option-set-value "general" "infobox") - it seems to decline any list value. Maybe the reason is deep in scheme (i.e. in scheme (eq? (list "a") (list "a")) isn't true whereas (equal? (list "a") (list "a")) is #t.
05:48:39 <chris> ... and looking through options.scm, multichoices have special handling that I can't decode
05:49:48 <chris> so, instead of trying to fix options.scm I think it'd be more sane to fix transaction.scm to store the values as simple 'symbols instead of a "clever" (list 'symbols or #chars)
05:50:09 <chris> ... while unstable is still cooking
06:08:13 *** ah has joined #gnucash
06:17:09 *** Aussie_matt has joined #gnucash
06:32:39 *** kus has joined #gnucash
07:05:40 *** User_ has joined #gnucash
07:11:51 *** User_ has quit IRC
07:33:33 *** bob_ has joined #gnucash
07:37:39 *** bob_ has quit IRC
07:38:19 *** Jimraehl1 has joined #gnucash
07:38:27 *** bob_ has joined #gnucash
07:40:08 *** Jimraehl1 has left #gnucash
07:44:39 *** kus has quit IRC
08:01:23 *** bob_ has quit IRC
08:25:01 *** Aussie_matt has quit IRC
09:01:30 *** Mechtilde has quit IRC
09:04:04 *** oozer has joined #gnucash
09:06:58 *** badger92 has quit IRC
09:14:02 *** chris has quit IRC
09:14:10 *** Mechtilde has joined #gnucash
09:18:33 *** oozer has quit IRC
09:20:47 *** frakturfreak has joined #gnucash
09:21:17 *** frakturfreak has quit IRC
09:52:27 *** chris has joined #gnucash
11:02:45 *** fabior has joined #gnucash
11:38:48 *** ah has quit IRC
11:49:21 *** bob_ has joined #gnucash
12:13:02 *** kus has joined #gnucash
12:13:36 <jralls> chris: I agree. Options is a Frankenstein's monster of C and Scheme that is nearly impossible to reason about.
12:27:11 <jralls> Mechtilde: Have you tried to build GnuCash from source? It's not hard on Debian, you can use apt-get build-dep gnucash for most stuff then add libwebkti2gtk-4.0-dev, libboost-all-dev, and googletest to pick up the remaining dependencies.
12:27:59 *** fabior has quit IRC
12:28:36 <Mechtilde> no I didn't try
12:30:20 <jralls> Can you, please? The transaction-delete crash should be fixed and if you build from source you can get debug symbols which will make it much easier to figure out the AQBanking one.
13:14:25 <gjanssens_afk> jralls: are you working on the removal of intltools ?
13:14:28 *** gjanssens_afk is now known as gjanssens
13:16:28 *** woodrec has joined #gnucash
13:20:48 <jralls> gjanssens: Not atm, I'm trying an optimization of the GncDateTime string constructor after profiling showed that it's a hotspot when loading a large database.
13:21:09 <jralls> gjanssens: After that I'll make the release.
13:21:15 <gjanssens> Ok
13:21:39 <jralls> BTW I'll be away next week.
13:22:08 <gjanssens> I am currently still working on migrating from .gnucash to .local/share/gnucash and .config/gnucash
13:22:25 <gjanssens> I believe on linux log.conf and config.user should be moved to the latter
13:22:49 <jralls> OK.
13:23:05 <gjanssens> In the process I started experimenting with boost::locale::translate for c++ translatable strings
13:23:24 <gjanssens> Unfortunately our current pot file generation code doesn't pick these up
13:23:35 <gjanssens> Only old style gettext (and _ )
13:23:52 <gjanssens> On the topic of being away: I'll be away from March 12 to March 26
13:24:40 <gjanssens> I had hoped to get the few extra translations into this weekend's release, but I may not make it in time.
13:24:51 <jralls> Doesn't boost::locale use ICU? That's a completely different structure. I don't think it's mixable with gettext.
13:25:09 <jralls> Translations? For nl_NL?
13:28:42 *** woodrec has quit IRC
13:29:17 <gjanssens> http://www.boost.org/doc/libs/1_56_0/libs/locale/doc/html/messages_formatting.html
13:29:49 <gjanssens> That page claims ordinary xgettext should be able to extract the strings if the proper keywords are added.
13:30:44 *** woodrec has joined #gnucash
13:35:48 *** woodrec has quit IRC
13:37:05 <jralls> Right, intltool knows only about the gettext() function. Notice that they're passing --keyword=translate to xgettext to get it to pick up their special function name.
13:39:20 <jralls> My optimization got a 33% reduction in load time, from 1.5 minutes to 1. \o/
13:40:44 <gjanssens> Nice :)
13:40:58 <gjanssens> I am trying to pass --keyword=translate as well
13:41:44 <gjanssens> I have added it to po/Makevars, but I'm starting to believe this file is no longer used in the cmake world
13:42:18 <gjanssens> Although intltool-update is trying to read it...
13:42:23 <gjanssens> Needs more debugging...
13:43:21 <jralls> I think you can use either intltool or xgettext to extract the messages but not both. You're using -G Makefiles, right?
13:43:50 <gjanssens> No, I'm on ninja
13:44:07 <gjanssens> And I understand it's either intltool or xgettext
13:44:15 <gjanssens> intltool does call xgettext btw
13:44:55 <jralls> Right, but it also needs to find one of Makefile, configure.ac, or configure.in in the root directory or it just quits.
13:45:22 <gjanssens> When I run xgettext by hand with the proper keywords, my strings are found and added to gnucash.pot
13:45:35 <jralls> More specifically intltool-update does. Dunno what intltool-sextract does.
13:45:36 <gjanssens> What quits ? intltool ?
13:45:59 <gjanssens> When I run make pot it does create a pot file, which is using intltool-update
13:46:52 <gjanssens> And there's no Makefile, configure.ac or configure.in in my source or build dir
13:47:34 <jralls> That's really odd. See CONF_Handle_Open around line 1081.
13:47:47 <jralls> Of intltool-update, that is.
13:48:06 *** badger92 has joined #gnucash
13:49:17 <jralls> Oh... It also looks for Makevars, so maybe that's why it doesn't die on you. Now I wonder why it was dying on me the other day.
13:49:18 <gjanssens> Oh, it's called Makevars, and that still exists.
13:51:06 <gjanssens> Huh, it should die because it can't find configure.ac or configure.in
13:51:22 <jralls> Hmm, but it still wants configure.ac or configure.in... but it's stupid and it will blindly set configure.in if it doesn't find configure.ac.
13:51:42 <gjanssens> Right, I just figured that as well...
13:52:34 <jralls> And then crash at line 1150 when it tries to open a nonexistent file.
13:53:34 <jralls> Which is not what it did for me the other day, it properly died with "Cannot find top_srcdir in Makefile".
14:03:40 <jralls> But I tried it just now and got "Unsuccessful open on filename containing newline at intltool-update line 1150.
14:08:23 <gjanssens> I see that as well, but that's a warning when generating POTFILES.in
14:08:48 <gjanssens> Found the issue with Makevars.
14:09:21 <gjanssens> We're running intltool-update --pot in the build directory, but Makevars is in the src dir.
14:09:34 <gjanssens> As intltool-update looks in the current directory, it won't find it
14:09:48 <gjanssens> I have added a configure_file to simply copy Makevars to the build dir and now it works
14:10:20 <gjanssens> The alternative is to run intltool-update in the source dir, but then gnucash.pot would be generated there as well.
14:10:47 <gjanssens> The better fix would be to drop intltool-update of course, but that's more than I have time for this evening
14:11:07 <gjanssens> If I can I'll pick that up next week.
14:16:13 <jralls> OK. Do you want your boost::locale changes in the release?
14:16:47 <gjanssens> I would like to yes.
14:16:55 <gjanssens> I'm almost done
14:17:58 <jralls> OK. I'll go for a bike ride, that'll take 1 1/2 hours or so, and start the release when I'm done.
14:21:16 <jralls> Hmm, we need some user entertainment for the data conversions on loading a 2.6 database. It can take several minutes on a big one.
14:29:25 *** sbluhm has quit IRC
14:32:48 *** woodrec has joined #gnucash