2022-05-28 GnuCash IRC logs
01:33:01 *** fell has quit IRC
01:34:21 *** fell has joined #gnucash
01:34:21 *** ChanServ sets mode: +o fell
02:36:28 *** storyjesse has joined #gnucash
02:45:02 *** sbluhm has joined #gnucash
02:53:24 *** Aussie_matt has quit IRC
02:55:08 *** Aussie_matt has joined #gnucash
03:07:45 *** fabior has joined #gnucash
03:17:17 *** sbluhm has quit IRC
03:17:29 *** storyjesse has quit IRC
03:17:34 *** Aussie_matt has quit IRC
03:18:10 *** storyjesse has joined #gnucash
03:18:11 *** Aussie_matt has joined #gnucash
03:27:03 *** sbluhm has joined #gnucash
03:40:33 *** sbluhm has quit IRC
03:46:56 *** sbluhm has joined #gnucash
03:59:37 *** Aussie_matt has quit IRC
04:00:27 *** sbluhm has quit IRC
04:04:27 *** Aussie_matt has joined #gnucash
04:23:26 *** storyjesse has quit IRC
04:24:42 *** storyjesse has joined #gnucash
04:25:28 <gjanssens> fell: where do you describe your gnucash-docs problem in more details (to save you from having to explain it all over again to me...)
04:25:28 <gncbot> gjanssens: Sent 2 days, 2 hours, and 48 minutes ago: <fell> New images from CWehli's doc PRs are included in FOP based formats, but not installed in GHELP or HTML.
04:25:57 <gjanssens> Ok there's the extra info :)
04:26:22 <gjanssens> I'll look into it. Just give me a few days.
04:26:24 <fell> Merge one of CWehli's PRs with new images
04:27:06 <fell> after you had already built the docs
04:27:52 <fell> On the new built the new images are ignored in xml and html
04:28:21 *** Aussie_matt has quit IRC
04:30:16 *** Aussie_matt has joined #gnucash
04:35:42 <fell> https://lists.gnucash.org/logs/2022/05/27.html#T14:11:31
04:39:41 *** sbluhm has joined #gnucash
05:13:28 *** Grav has joined #gnucash
05:14:40 *** ArtGravity has quit IRC
05:20:11 <gjanssens> fell: this is a weakness/flaw in out build scripts.
05:21:11 <gjanssens> In https://github.com/Gnucash/gnucash-docs/blob/maint/cmake/AddGncDocTargets.cmake#L5 we compose a list of all figures based on file globbing.
05:21:59 <fell> can we do it like in epub ?
05:22:14 <gjanssens> The CMake documentation warns this will not work very well when new files are added as the glob will only be performed during a cmake run and not during a make/build run
05:22:33 <gjanssens> That means that new figures aren't detected until you rerun cmake explicitly on the build directory.
05:22:43 *** jmdaemon has quit IRC
05:23:22 <fell> There we have "COMMAND cmake -E copy_directory "${CMAKE_CURRENT_SOURCE_DIR}/figures" "${EPUB_TMPDIR}/OEBPS/figures""
05:23:26 <gjanssens> This was a trade-off I made back then because the alternative is to explicitly list all figures manually in the cmake files somewhere. This was and still is a burden to documentation writers.
05:25:46 <gjanssens> I don't think doing for ghelp what we do for epub is ideal either.
05:26:42 <fell> Or trigger the Cmake, if one of the figures dirs change?
05:27:38 <gjanssens> The epub build is such that if any single file changes, the whole epub build must be rerun. That's why it by chance will always trigger a correct rebuild if we add a figure - adding a figure always goes together with changing an xml file to use that image. The xml file change triggers the build.
05:27:49 <gjanssens> A full build.
05:28:38 <gjanssens> For ghelp, changes to one xml file won't affect any other xml files, so adding a single figure in a single xml file should only copy the changed xml file and the new figure, not everything.
05:29:13 <gjanssens> As for your Cmake trigger, I don't think cmake can trigger on directory content changes, only on file changes.
05:30:39 <gjanssens> I'm inclined to forgo my trade-off and explicitly list all figures used in the */*/CMakeLists.txt files. Adding new images then means again adding it in the appropriate CMakeLists.txt file as well.
05:31:21 <gjanssens> I have had to do the same for our pixmaps in gnucash/data.
05:31:45 <gjanssens> It's how cmake is intended to be used and we should stick to that.
05:32:01 <gjanssens> Even it that's slightly more cumbersome for our documentation writers/translators.
05:32:44 <gjanssens> As devs we can also help them by adding the proper entries if the writer's PRs are lacking them.
05:34:38 <gjanssens> I mean we can to that at merge time on their behalf if they forgot.
05:34:45 <gjanssens> s/to/do/
05:36:20 <fell> I dont like that idea as we are adding/updating many images now.
05:38:32 <fell> If I update an image,which Ionce made under GTK2 to TK3, I will not neccessarily change the XML.
06:04:37 *** storyjesse1 has joined #gnucash
06:04:51 *** storyjesse has quit IRC
06:04:52 *** storyjesse1 is now known as storyjesse
06:13:17 *** fabior has quit IRC
06:18:40 *** fabior has joined #gnucash
06:43:11 *** fell has quit IRC
06:43:28 *** fell has joined #gnucash
06:43:28 *** ChanServ sets mode: +o fell
06:45:42 *** sbluhm has quit IRC
06:45:59 *** fabior has quit IRC
06:46:51 *** fabior has joined #gnucash
06:49:47 *** fabior has quit IRC
06:51:18 *** fabior has joined #gnucash
07:02:46 <gjanssens> fell: changing images is no issue. Cmake will properly detect that. It only fails to detect added images.
07:04:29 *** fabior has quit IRC
07:06:20 <fell> A quick approach for a list: 'ls -{1|m}'
07:06:39 <fell> smarter: an xslt
07:07:02 <gjanssens> None of those help cmake.
07:08:04 *** fabior has joined #gnucash
07:08:47 <gjanssens> To know when to rerun, cmake adds a few clever rules in the build system. That's based on dependency lists.
07:09:11 <gjanssens> You can't however not force the build system to regenerate the dependencies based on the contents of a directory.
07:09:24 <gjanssens> The build system can only test file timestamps.
07:09:49 <gjanssens> So a change of a CMakeLists.txt file (like adding a new image file) will trigger cmake to rerun.
07:11:04 *** sbluhm has joined #gnucash
07:11:17 <gjanssens> Having a rule that just lists all images programmatically won't cause such a timestamp change of a tracked file and hence cmake won't run if you just add an image in such scenario
07:11:53 <gjanssens> Whether that's due to running cmake's own file globbing logic or due to a shell script or an xsl transform doesn't matter.
07:12:46 <gjanssens> If you don't like a manual list of image files, your other option is to remember to force a cmake run (perhaps with the option to ignore the cache)
07:13:09 <gjanssens> But then we also need a mechanism to trigger this in the automated nightly builds and tests.
07:13:17 <gjanssens> (ci tests)
07:13:51 *** fabior has quit IRC
07:15:59 <gjanssens> By the way I notice a figures100 directory in guide/pt ? Why has that been added ?
07:18:46 <fell> Oh, Pedro had some issues with git like merge insted of rebase …
07:20:00 <fell> and John was alone at release time.
07:36:19 *** fabior has joined #gnucash
07:45:15 *** fabior has quit IRC
07:48:37 *** sbluhm has quit IRC
07:56:10 *** sbluhm has joined #gnucash
08:05:50 <gjanssens> Are they meant to be removed again then ?
08:10:26 *** sbluhm has quit IRC
08:19:22 *** sbluhm has joined #gnucash
08:26:02 *** georgepletsas has joined #gnucash
08:28:23 *** sbluhm has quit IRC
08:30:43 *** georgepletsas has quit IRC
10:40:16 *** storyjesse has quit IRC
10:48:55 *** sbluhm has joined #gnucash
10:55:48 *** Aussie_matt has quit IRC
11:10:39 *** sbluhm has quit IRC
11:36:35 *** sbluhm has joined #gnucash
11:42:36 *** sbluhm has quit IRC
11:56:05 *** chris has joined #gnucash
11:56:06 *** ChanServ sets mode: +v chris
11:56:09 *** gncbot sets mode: +o chris
12:12:16 *** AE has joined #gnucash
12:19:38 *** sbluhm has joined #gnucash
12:24:29 *** AE is now known as Viaduct
12:25:12 *** ChanServ sets mode: +v Viaduct
12:26:35 <Viaduct> I'm after some help with hacking custom reports. I have a dashboard setup with a number of asset charts for various account views. It would be really nice if I could hack the scheme code to have an end date of 'yesterday' in the charts rather than 'today' since the value of my equities are based on yesterdays close.
12:26:50 <Viaduct> There is a line in the scheme report code for each chart for (relative . Today). Can this be hacked to show yesterday? I tried (relative . yesterday) but it didn't have that defined.
12:26:59 <Viaduct> Any help appreciated if any scheme experts are online? Thanks
12:29:10 *** Viaduct has left #gnucash
12:29:27 *** Viaduct has joined #gnucash
12:29:27 *** ChanServ sets mode: +v Viaduct
12:29:52 <Viaduct> I'm after some help with hacking custom reports. I have a dashboard setup with a number of asset charts for various account views. It would be really nice if I could hack the scheme code to have an end date of 'yesterday' in the charts rather than 'today' since the value of my equities are based on yesterdays close.
12:30:06 <Viaduct> There is a line in the scheme report code for each chart for (relative . Today). Can this be hacked to show yesterday? I tried (relative . yesterday) but it didn't have that defined.
12:30:16 <Viaduct> Any help appreciated if any scheme experts are online? Thanks
12:32:13 <Viaduct> Oops. Apologies, may have posted twice. New to IRC. 🙄
12:38:54 *** Viaduct has quit IRC
12:43:55 *** AE has joined #gnucash
12:48:40 *** AE has quit IRC
12:50:15 *** Viaduct has joined #gnucash
12:50:15 *** ChanServ sets mode: +v Viaduct
12:53:22 *** Viaduct has quit IRC
12:53:48 *** Viaduct has joined #gnucash
12:55:28 *** Viaduct has quit IRC
13:03:54 *** Viaduct has joined #gnucash
13:16:38 *** sbluhm has quit IRC
13:19:58 *** sbluhm has joined #gnucash
13:42:45 *** Viaduct has joined #gnucash
13:44:20 *** Viaduct has quit IRC
13:46:16 *** sbluhm has quit IRC
14:01:39 *** gjanssens has quit IRC
14:05:07 *** dckc has joined #gnucash
14:18:25 *** sbluhm has joined #gnucash
14:46:33 *** dckc has quit IRC
15:34:56 *** sbluhm has quit IRC
15:45:52 *** sbluhm has joined #gnucash
17:03:08 *** sbluhm has quit IRC
19:11:26 *** phebus has quit IRC
20:54:54 *** storyjesse has joined #gnucash
22:04:20 *** fell has quit IRC
22:05:09 *** fell has joined #gnucash
22:05:09 *** ChanServ sets mode: +o fell