2018-09-25 GnuCash IRC logs

00:39:03 *** jethrob has quit IRC
00:41:19 *** jethrob has joined #gnucash
00:51:28 *** Mechtilde has joined #gnucash
01:27:04 *** Cuare has joined #gnucash
01:51:14 *** Mechtilde has quit IRC
02:46:09 *** gjanssens has joined #gnucash
02:46:09 *** ChanServ sets mode: +o gjanssens
02:46:41 <gjanssens> .
02:58:19 *** fabior has joined #gnucash
03:49:45 *** jotrago has quit IRC
04:06:05 *** bertbob has quit IRC
04:11:57 *** bertbob has joined #gnucash
04:15:08 *** ncv has joined #gnucash
04:55:36 *** Mechtilde has joined #gnucash
04:58:34 *** Cuare has quit IRC
05:02:23 *** Cuare has joined #gnucash
05:06:53 *** storyjesse has joined #gnucash
05:09:22 *** storyjesse has quit IRC
05:15:24 *** puck has quit IRC
05:23:57 *** puck has joined #gnucash
06:04:49 *** pilotauto_ has quit IRC
06:35:09 *** oozer has joined #gnucash
06:44:26 *** Mechtilde has quit IRC
06:45:39 *** fabior has quit IRC
06:47:02 *** fabior has joined #gnucash
06:49:52 *** fabior has quit IRC
07:12:20 *** Jimraehl1 has joined #gnucash
07:13:06 *** Jimraehl1 has left #gnucash
08:00:57 *** Mechtilde has joined #gnucash
08:01:22 *** KevinDB has quit IRC
08:05:16 *** KevinDB has joined #gnucash
09:22:43 *** chris has quit IRC
09:24:53 *** chris has joined #gnucash
09:27:09 *** fabior has joined #gnucash
09:46:27 *** jotrago has joined #gnucash
10:00:56 *** jotrago has quit IRC
10:24:24 *** kael has joined #gnucash
10:38:07 *** KaiForce has quit IRC
10:54:28 <warlord> .
11:15:13 *** Mechtilde has quit IRC
11:28:02 *** jotrago has joined #gnucash
11:41:16 *** frakturfreak has joined #gnucash
11:50:03 *** fabior has quit IRC
11:58:49 *** fabior has joined #gnucash
12:14:08 <jralls> gjanssens: wrt 9c9068cfa, we need to think about GtkAction: It's deprecated since Gtk3.10, no telling when they'll decide to drop it.
12:15:02 <gjanssens> jralls: I noticed. The proper replacement is GAction and family.
12:15:25 <gjanssens> It would have been too big a project to do in this stabilization effort though
12:15:27 <jralls> Does that require us to adopt GMenuModel as well?
12:16:16 <gjanssens> I don't know. I haven't dug much deeper yet.
12:16:17 *** frakturfreak has quit IRC
12:16:20 <gjanssens> Would that be an issue ?
12:17:46 <gjanssens> It does look like we should adopt GMenuModel as well: https://developer.gnome.org/gtk3/stable/GtkUIManager.html has a more detailed deprecation notice
12:22:32 <jralls> The Gtk side of GMenuModel takes care of Mac menu integration, but via GApplication--or more specifically, GtkApplicationQuartzImpl, so another layer. It will be a bit of work.
12:23:26 <gjanssens> That's what I feared.
12:24:07 <gjanssens> I don't expect gtk to drop the deprecations in gtk3 though, but one never knows...
12:24:38 <gjanssens> ISTR Bob looked at this migration in the passed and backed off precisely because of the large impact.
12:24:44 <jralls> They won't. Gtk3 is in maintenance mode, any API changes will go to Gtk4.
12:25:45 <gjanssens> That means we can plan the required changes for gnucash 4 as well IMO
12:26:17 <jralls> Even when they actually get Gtk4 stable enough to release (no sign of that yet) Gtk3 will still be in distros for a while. The main worry will be webkit stuffing us again.
12:27:29 <gjanssens> Indeed. *That* will require rethinking for gnucash 4
12:27:29 <jralls> So I'd be inclined to put more effort into getting rid of webkit in GC4 rather than worrying too much about Gtk4 menus and builder vs uimanager.
12:27:42 <gjanssens> Agreed
12:30:10 *** fabior has quit IRC
12:32:29 *** frakturfreak has joined #gnucash
12:40:09 *** fabior has joined #gnucash
12:41:20 *** frakturfreak has quit IRC
12:42:37 *** frakturfreak has joined #gnucash
12:44:35 *** calvinct has joined #gnucash
12:58:27 *** Mechtilde has joined #gnucash
12:59:19 *** kus has joined #gnucash
13:08:41 *** calvinct has quit IRC
13:12:14 *** fabior has quit IRC
13:12:31 *** frakturfreak has quit IRC
13:15:16 *** ncv has quit IRC
13:56:21 *** calvinct has joined #gnucash
14:18:29 *** calvinct has quit IRC
14:32:02 *** frakturfreak has joined #gnucash
14:38:31 *** kael has quit IRC
14:40:13 *** kael has joined #gnucash
14:48:30 *** frakturfreak has quit IRC
14:57:24 *** frakturfreak has joined #gnucash
15:00:30 *** frakturfreak has quit IRC
15:14:44 *** frakturfreak has joined #gnucash
15:16:24 *** Mechtilde has quit IRC
15:24:57 *** kael has quit IRC
15:46:15 <gjanssens> jralls: I have just discovered we have different release tarballs on github and sourceforge - their checksums differ
15:46:46 <gjanssens> Looking closer it appears this is due to different generation date in the included gnucash.pot file
15:47:12 <gjanssens> gnucash-3.2.tar.bz2 on sourceforge matches the published checksum
15:47:19 <gjanssens> The one on github doesn't
15:47:33 <gjanssens> Do you remember uploading from different tarballs ?
15:49:06 <gjanssens> I ran into this because flatpak requires a checksum for builds from tar archives and I found the checksum we published not matching the checkum of the associated tarball
15:52:05 *** kael has joined #gnucash
16:08:51 <chris> quick qu - if webkit is going in gc4, how will reports be displayed? :-/
16:12:19 <chris> the only way i can see things happen in this direction is by directing report html output to an external web browser, and open a port locally to accept links to register/accounts/reports
16:13:39 <chris> https://localhost:8081/gnc-register:acct-guid=58c725be349d47d2b8cbc5f12ade7b84#
16:18:33 <gjanssens> chris: webkit is not the only html rendere out there
16:18:39 <gjanssens> renderer*
16:18:44 <chris> ohh ok
16:19:07 <gjanssens> But the exact details are not known yet
16:19:57 <gjanssens> One option would be to move to a native webengine on OS X and Windows.
16:20:07 <gjanssens> Probably not the preferred solution, but possible
16:20:59 <gjanssens> I believe Windows does have an api to embed their edge web engine (or was it the older internet explorer, I don't remember the details)
16:21:40 <gjanssens> But as jralls has rightfully stated several times in the past, a complete web engine is serious overkill for our limited needs
16:22:01 *** jerryq has joined #gnucash
16:22:06 <gjanssens> So we may search for a lighter alternative when the time is right
16:22:22 <chris> sure but whatever replaces it must understand tables and javascript
16:22:43 <gjanssens> yes, I have said the same thing several times in the past.
16:23:06 <chris> i've been able to experiment running a guile webserver by creating a report with threads, it can accept requests from external browser, but this would open floodgates to wishes for a full web interface
16:23:10 <gjanssens> jralls: I have worked some more on the flatpak package generation scripts
16:23:57 <gjanssens> They can now build from tarball and will do so automatically if the revision to build is a tag
16:24:26 <gjanssens> The code assumes all tags are releases.
16:24:41 <gjanssens> This has been the case for as long as we're using git.
16:25:10 <jralls> gjanssens, chris: The native web code on MacOS is webkit. The issue is getting it onto a Gtk window. That would also be a problem with Edge, plus we'd have to code an html-edge.c to go along with html-webkit.c
16:26:01 <chris> ok... and if we're supporting windows7 it'd be html-ie9 or -trident something
16:26:02 <gjanssens> jralls: indeed. I think I got this idea from wxwidgets
16:26:27 <gjanssens> But that is written to use native api's under the hood of course
16:26:32 <jralls> Yes, I think that's the way they do it.
16:28:10 <gjanssens> The other option we discussed in the past was a combination of a simple html rendering library with a javascript engine
16:28:35 <jralls> As for tarballs, they should be the same, I do make dist then upload the results to both sites. But on github you have to be careful to get https://github.com/Gnucash/gnucash/releases/download/3.2/gnucash-3.2.tar.gz and not https://github.com/Gnucash/gnucash/archive/3.2.tar.gz.
16:30:03 <gjanssens> jralls: go check for yourself: they are not
16:30:04 <jralls> Hmm, but I see that it's not the same.
16:30:36 <gjanssens> They must be from different runs of make/ninja dist
16:30:57 <gjanssens> Probably on the same release tag.
16:31:19 <gjanssens> comparing the source trees the only difference is the timestamp in po/gnucash.pot
16:31:45 <gjanssens> Which is set when the pot file is generated
16:32:07 <gjanssens> This normally only happens on make/ninja dist or when explicitly calling make po/gnucash.po
16:35:20 <gjanssens> gnucash.pot in the github tarball is about 5 hours more recent than the on in the sf tarball
16:36:59 <gjanssens> Perhaps we should replace this time stamp with the timestamp of the git commit from which make dist was run to avoid such surprises in the future ?
16:37:26 <gjanssens> It's a bit silly that generating a dist tarball from exactly the same commit would result in different tarballs
16:37:28 *** calvinct has joined #gnucash
16:38:06 <gjanssens> I know it's only a timestamp but users don't know this if they try to download our tarball and verify the checksum
16:39:59 <gjanssens> We already capture date information for other information strings in the build so that info should be reusable.
16:42:22 <jralls> You might be overthinking this. If the only difference is the timestamp in gnucash.pot then the obvious thing to do is to copy the sourceforge one to github and be done with it.
16:43:05 <jralls> I can't think of a good reason that I would have rebuilt for one and not the other.
16:43:35 <jralls> And obviously if that's the only difference there wasn't any reason to have rebuilt at all.
16:44:18 <gjanssens> Of course copying the tarball is the immediate solution. I was more thinking long-term
16:44:59 <gjanssens> But it's probably not that important
16:45:42 <gjanssens> For the flatpak builds I have chosen to stick with sourceforge to download our sources
16:46:20 <gjanssens> The primary reason is that we have a README.txt file there from which I can automatically extract the requires sha256 checksums
16:46:40 <gjanssens> On github this is embedded in the news page which is a bit harder to parse in shell script
16:47:06 <gjanssens> Can we add an explicit checksum file with each release as an artefact there as well ?
16:47:25 <gjanssens> (a downloadable artefact that is)
16:50:44 <jralls> Sure.
16:52:09 <gjanssens> Ok. Though that's only half of the issue. We don't publish our gnucash-docs dist tarball on github so I'll have to use sf anyway.
16:52:42 *** calvinct has quit IRC
16:53:07 <gjanssens> The release tab on gnucash-docs offers a github source tarball which is not the same as our dist tarball IIUC
16:53:21 <gjanssens> It's probably missing the generated configure and Makefile.in files
16:54:00 <gjanssens> It would be good to convert gnucash-docs to cmake as well, which would eliminate this issue.
16:54:21 <jralls> That's correct. The dist-generated tarball is attached to the GnuCash release.
16:54:28 <gjanssens> In that case there would probably be no difference any more between a source and dist tarball for docs
16:54:32 <jralls> https://github.com/Gnucash/gnucash/releases/download/3.2/gnucash-docs-3.2.tar.gz
16:54:44 <gjanssens> Oh, that's good enough
16:55:14 <gjanssens> Which brings me to another oddity I noticed while searching for dist tarballs
16:55:48 <gjanssens> On github the download links are pretty consistent. They all start with https://github.com/Gnucash/gnucash/releases/download/<rel>/
16:56:16 <gjanssens> On sf we have separate directories for gnucash and gnucash-docs
16:56:29 <gjanssens> Probably that grew historically so
16:57:18 <gjanssens> I was tempted to restructure this for newer releases to start with a <ver>/directory and dumping all build artefacts in there
16:57:28 <jralls> Yes, we only started releasing docs and gnucash together around halfway through 2.6. Before that docs were released rather haphazardly.
16:57:32 <gjanssens> That is stop using a different subdirectory for gnucash-docs
16:58:27 <jralls> I don't have a problem with that. Do you have shell access set up on SF? It will be a lot easier from there than in the web interface.
16:58:48 <gjanssens> I would have to check, I think so
17:01:29 <jralls> Looking at gnucash.pot, we should fix the top comment. "SOME DESCRIPTIVE TITLE.\nCopyright (C) YEAR THE PACKAGE'S COPYRIGHT HOLDER" looks pretty dumb.
17:07:32 <gjanssens> Indeed.
17:07:53 <gjanssens> I do have shell access, I will play with restructuring the directories later
17:08:08 <gjanssens> Now it's time to wrap up for this evening
17:08:23 <jralls> Good night!
17:10:44 <gjanssens> Bye
17:12:34 *** gjanssens has quit IRC
17:51:46 *** calvinct has joined #gnucash
17:53:33 *** calvinct has quit IRC
17:54:09 *** calvinct has joined #gnucash
17:57:38 *** calvinct has quit IRC
18:09:09 *** dad has joined #gnucash
18:10:51 <dad> how do you connect gnucash-docs to the gnucash program? I am using 3.2 and both program and docs are located in separate directories under my Home
18:12:15 *** dad has quit IRC
18:12:30 *** dad has joined #gnucash
18:13:39 *** dad has quit IRC
18:14:58 *** dad has joined #gnucash
18:20:23 *** dad has joined #gnucash
18:28:29 *** jerryq has quit IRC
18:32:21 *** kael has quit IRC
18:35:34 *** pilotauto has joined #gnucash
19:11:07 *** jerryq has joined #gnucash
19:11:36 *** josharmenta has joined #gnucash
19:14:07 *** josharmenta has quit IRC
19:30:07 *** dad has quit IRC
19:38:31 *** dad has joined #gnucash
19:44:42 <dad> I compiled gnucash 3.2 on an ubuntu derivative (Pop_OS), but there is no help in the program. I compiled gnucash-docs in a separate directory as recommended but now I don't know how to integrate the docs into the program. Both directories are under my Home. Any help would be much appreciated.
19:51:41 *** CDB-Man_ has joined #gnucash
19:52:15 *** CDB-Away_ has joined #gnucash
19:52:56 *** CDB-Away has quit IRC
19:53:30 *** CDB-Man has quit IRC
20:03:35 <jralls> dad: They need to go in x/share/doc/gnucash-docs where x is in $XDG_DATA_DIRS.
20:06:19 <jralls> dad: The easiest way to go is to set --prefix (for docs) and -DCMAKE_INSTALL_PREFIX (for GnuCash) to $HOME/.local and do make install in both gnucash-build and gnucash-docs-build. Make sure XDG_DATA_DIRS includes $HOME/.local/share and it should work.
20:14:13 <dad> jralls: Thanks I'll try your suggestions.
20:26:23 *** dad has quit IRC
20:57:18 *** calvinct has joined #gnucash
20:59:47 *** dad has joined #gnucash
21:00:51 <dad> jralls: Thanks it worked.
21:01:10 *** dad has quit IRC
21:18:05 *** calvinct has quit IRC
21:24:56 *** oozer has quit IRC
23:45:41 *** marusich has joined #gnucash
23:53:18 *** badger93 has joined #gnucash
23:55:45 *** badger92 has quit IRC
23:55:45 *** badger93 is now known as badger92