2018-01-25 GnuCash IRC logs

00:15:55 *** jethrogb has quit IRC
00:34:38 *** jethrogb has joined #gnucash
01:00:28 *** jotrago has quit IRC
01:20:35 *** Mechtilde has joined #gnucash
01:31:54 *** fell has quit IRC
01:33:01 *** fell has joined #gnucash
01:57:51 *** marusich has quit IRC
01:58:10 *** marusich has joined #gnucash
02:03:52 *** marusich has quit IRC
02:04:10 *** marusich has joined #gnucash
02:08:25 *** Mechtilde has quit IRC
02:11:34 *** pilotauto has quit IRC
02:32:56 *** fekepp has quit IRC
02:58:09 *** fekepp has joined #gnucash
03:00:00 *** gjanssens has joined #gnucash
03:00:00 *** ChanServ sets mode: +o gjanssens
03:00:17 <gjanssens> .
03:20:47 *** bob_ has joined #gnucash
03:23:52 *** marusich has quit IRC
03:24:09 *** marusich has joined #gnucash
03:27:45 *** marusich has quit IRC
03:28:03 *** marusich has joined #gnucash
03:42:43 *** bob_ has quit IRC
04:00:32 *** fekepp has quit IRC
04:01:29 *** fekepp has joined #gnucash
04:13:11 *** rubdos has joined #gnucash
04:18:43 *** harshitaneja has joined #gnucash
04:21:16 *** meb has joined #gnucash
04:25:22 *** meb has quit IRC
04:33:23 *** jotrago has joined #gnucash
04:57:56 *** fabior has joined #gnucash
05:04:12 *** gncbot has joined #gnucash
06:10:09 *** jotrago has quit IRC
06:30:31 *** User has joined #gnucash
06:33:13 *** oozer has joined #gnucash
06:33:46 *** User has quit IRC
07:09:47 *** fabior has quit IRC
07:21:42 *** kus has joined #gnucash
07:34:03 *** kus has quit IRC
09:56:04 *** gncbot has joined #gnucash
09:56:11 *** ChanServ sets mode: +qo warlord warlord
09:56:13 *** warlord sets mode: +o gncbot
09:56:23 <warlord> so finicky!
09:56:23 <gncbot> warlord: Sent 1 day, 4 hours, and 4 minutes ago: <chris> or anyone - been checking devel archives... from your memory - when running Tools>Close Book, does it always create transaction with xaccTransGetIsClosingTxn = true? if so, then the Closing Entries pattern should be removed from equity-statement.scm, trial-balance.scm, income-statement.scm and converted to a boolean filter?
09:57:42 <warlord> @tell chris I don't think it ALWAYS did that, unfortunately. I think that boolean was added later. We could attempt to build a filter that would add that flag on file conversion from an older version going forward?
09:57:42 <gncbot> warlord: The operation succeeded.
09:58:03 <Malte[m]> > hiho, i've my transaction records in mt940 but with some gaps - but was able to pull some saldi from paper records. is it possible to combine these to find remaining gaps and calculate transactions without record?
09:59:01 <warlord> Malte[m]: perhaps by hand -- or using the reconcile function?
09:59:49 <Malte[m]> hm
10:00:39 <Malte[m]> i'll try
10:02:25 <warlord> gjanssens: for some reason the docs build has been erroring out with "git checkout failed". Alas, I don't have any additional info (yet), but I've no clue why that would happen. I;m not sure if it's the doxygen or gnucash-docs build failing. I'll go research.
10:25:08 *** Mechtilde has joined #gnucash
10:35:35 *** fekepp has quit IRC
10:40:33 *** kael has joined #gnucash
10:45:36 *** wol has joined #gnucash
10:54:09 <wol> Hallo Mechtilde, die Aktivitäten hier sind wohl eher gering?
10:55:22 <Mechtilde> das kommt auf die Zeit an
10:55:27 *** kael1 has joined #gnucash
10:58:13 *** kael has quit IRC
10:58:19 *** kael1 is now known as kael
10:59:01 *** harshitaneja has quit IRC
11:08:06 *** Cuare has joined #gnucash
11:25:53 *** kael has quit IRC
11:33:48 *** kael has joined #gnucash
11:35:16 *** kael has quit IRC
11:35:30 *** kael has joined #gnucash
11:39:57 *** kael has quit IRC
11:44:55 *** kael has joined #gnucash
11:45:29 *** Mechtilde has quit IRC
11:58:32 *** Mechtilde has joined #gnucash
11:58:41 *** kael has quit IRC
11:58:53 *** kael has joined #gnucash
12:00:46 *** kael has quit IRC
12:04:42 *** gncbot sets mode: +o fell
12:08:53 <fell> gjanssens: You finished RL accounting?
12:09:43 *** harshitaneja has joined #gnucash
12:12:24 *** ArtGravity has joined #gnucash
12:13:33 *** wol has quit IRC
12:14:44 *** wol has joined #gnucash
12:22:56 *** harshitaneja has quit IRC
12:23:39 *** harshitaneja has joined #gnucash
12:26:27 *** jralls has joined #gnucash
12:26:27 *** ChanServ sets mode: +o jralls
12:27:26 <jralls> .
12:32:08 <fell> warlord: https://code.gnucash.org/docs/MASTER/ is empty
12:35:45 <jralls> That's the *other* docs build. It probably depends on autotools.
12:37:45 <luc14n0> Can somebody confirm that cmake is not generating symlinks? The travisci unstable branch's cmake+ninja (the one I'm interested in) raw log have encoding issues and I can't figure out if there is any *.so.?.?.?
12:39:49 <jralls> Looks like the main docs build ran this morning, though I don't see why it would; the last commit was 4 days ago.
12:40:43 <jralls> luc14n0: I'm pretty sure the CMake build isn't set up to make any links on libgnc-foo. Why would they be needed?
12:41:27 <jralls> luc14n0: I don't think that the autotools build did either...
12:41:58 <luc14n0> jralls: autotools used to produce it
12:42:25 <warlord> fell: the doxygen build definitely depends on autotools.
12:42:42 <warlord> but that doesn't explain the "git checkout failed" error I'm seeing.
12:43:10 <warlord> the docs build runs every night; it doesn't check for updates.
12:44:31 <warlord> jralls: this is the line that is failing: git checkout $branch -q >/dev/null 2>&1 || my_die "git checkout failed."
12:44:37 <jralls> warlord: Hmm, something's screwy then: It updated the help epub, mobi, and pdf but not the guide. Both html dirs updated.
12:44:49 <warlord> Where $branch == master
12:45:39 <warlord> I'm only guessing whether it's the doxygen or gnucash-docs that's failing. Let me check
12:46:44 <warlord> Okay, it's definitely the doxygen build that's failing with git checkout failed.
12:47:03 <jralls> warlord: It's doxygen. You'll have to run the checkout by hand to see the real error; you might consider sending stderr to a variable in the script and echoing it in the "die" message.
12:47:58 <warlord> That would be smart. ;)
12:48:19 <warlord> Anyways, the gnucash-docs build should be working, provided it still uses autotools.
12:50:12 <jralls> Like I said, the gnucash-docs build seems to have run just fine this AM, aside from the weirdness about updating the maint e-docs and not the master ones.
12:50:32 <jralls> Sigh, no, I mean the help e-docs and not the guide ones.
12:51:08 <warlord> bash-4.3$ git checkout master
12:51:08 <warlord> fatal: Not a git repository (or any parent up to mount point /home)
12:51:08 <warlord> Stopping at filesystem boundary (GIT_DISCOVERY_ACROSS_FILESYSTEM not set).
12:51:34 <jralls> So the directory got deleted.
12:52:03 <warlord> the gnucash-docs only builds maint.
12:52:17 <warlord> doxygen builds both maint and master.
12:52:20 <jralls> Yeah, see the "Sigh, no..."
12:52:32 <fell> IMHO it is right to build maint of the docs
12:52:57 <warlord> w.r.t. the git checkout, the directory *does* exist. I can cd into it. It's a full tree. There's a .git/config file in there..
12:52:58 <jralls> Of course it is. That's the only branch that gets any changes.
12:53:23 <jralls> Ah, you have to be cd'd into it in order to run git.
12:53:28 *** harshitaneja has quit IRC
12:53:48 *** harshitaneja_ has joined #gnucash
12:54:04 <jralls> And so does the script.
12:54:10 <warlord> Yes, it does. I am..
12:54:30 <warlord> I think *this* might be the problem:
12:54:31 <warlord> bash-4.3$ ls -lta .git/
12:54:31 <warlord> total 60
12:54:31 <warlord> drwxrwxr-x. 8 gnucash-docs gnucash-docs 198 Jan 21 01:00 .
12:54:31 <warlord> -rw-r--r--. 1 gnucash-docs gnucash-docs 0 Jan 21 01:00 ORIG_HEAD
12:54:31 <warlord> -rw-r--r--. 1 gnucash-docs gnucash-docs 0 Jan 21 01:00 index
12:54:31 <warlord> -rw-rw-r--. 1 gnucash-docs gnucash-docs 0 Jan 21 01:00 FETCH_HEAD
12:54:33 <warlord> -rw-r--r--. 1 gnucash-docs gnucash-docs 0 Jan 21 01:00 HEAD
12:55:18 *** harshitaneja_ is now known as harshitaneja
12:58:01 <fell> files should only have -rw-r--r--
12:58:41 <jralls> Yes, that would do it. index should be 298650, HEAD 23, ORIG_HEAD and FETCH_HEAD would depend on the last successful operation.
12:59:12 <jralls> The stray group write priv isn't the problem: The empty index is.
12:59:37 <jralls> Prolly ths simplest thing to do is rm -r and re-clone.
13:00:31 <warlord> Similar issue in maint :(
13:00:52 <warlord> Then I need to fix the autotools dependency
13:03:48 *** kael has joined #gnucash
13:05:01 <warlord> jralls: Can you help me update this line:
13:05:04 <warlord> # Compute the version from the configure script
13:05:04 <warlord> VERSION=`grep AC_INIT configure.ac | awk '{print $2}' | sed -e 's/\[//' -e '
13:05:04 <warlord> s/\],//'`
13:06:38 <jralls> warlord: How about "`util/gnc-vcs-info -r .`"?
13:09:59 <warlord> jralls: that gives the VCS ID, not the version number. I want the version number (e.g. 2.6.18)
13:12:08 *** wol has quit IRC
13:12:28 <warlord> Anyways, a re-clone fixed the immediate problem.
13:13:09 <warlord> It looks like there is still a configure.ac in master?
13:13:53 <jralls> warlord: Yes, because unstable hasn't been merged up. I'm supposed to release 2.7.4 this weekend and I'll merge it up then.
13:16:09 <jralls> gnc-vcs-info returns 2.7.3-1-g310442f, the first bit of which is the current tagged version number for the branch. Since we bump the version in CMakeLists.txt and tag that commit with the version it should be the same. Just pass the result through a regex to get the plain version number if that's what you want.
13:17:52 <jralls> If you'd rather grep the CMakeLists.txt, `grep PACKAGE_VERSION CMakeLists.txt` will give you "SET (PACKAGE_VERSION 2.7.3)" which you can easily filter with a regex.
13:22:02 *** kael has quit IRC
13:27:39 *** wol1 has joined #gnucash
13:28:20 <warlord> Hmm, I missed that. I can use that.
13:29:21 <warlord> Ah, maint's version doesn't give that. BUT, I can base it on the existence of configure.ac.
13:33:38 <jralls> luc14n0: Sorry for the delay, I had to build with autotools to see what it's doing. Instead of just creating libgncmod-foo.so like cmake does it creates libgncmod-foo.so.0.0.0 and then links libgncmod-foo.so.0 to that and libgncmod-foo.so to libgncmod-foo.so.0. Utterly useless since we don't yet version our implementation libraries.
13:34:42 <warlord> Okay, updated the script:
13:34:55 <warlord> # Compute the version from the configure script
13:34:55 <warlord> if [ -f configure.ac ] ; then
13:34:55 <warlord> VERSION=`grep AC_INIT configure.ac | awk '{print $2}' | sed -e 's/\[//'\
13:34:55 <warlord> -e 's/\],//'`
13:34:55 <warlord> else
13:34:56 <warlord> VERSION=`util/gnc-vcs-info -r .`
13:34:57 <warlord> fi
13:38:42 <jralls> Nice.
13:39:13 <jralls> Come to think of it I don't remember ever testing doxygen + cmake. I'll do that now.
13:39:41 <warlord> Oh... Hmm.. it also depends on the doc/Makefile.am
13:40:07 <warlord> Is that going away, too?
13:40:28 <warlord> (is there any chance it could not go away?)
13:42:35 <fell> Too late, warlord, you did not support me when I complained. ;-)
13:43:33 <jralls> Heh, no doxygen at all. A project for today, I guess.
13:44:08 <jralls> But no, doc/Makefile.am won't do any good without the rest of autotools.
13:44:23 <warlord> jralls: sure it will.
13:44:27 <warlord> At least for me!
13:44:31 <jralls> How?
13:44:42 <warlord> My code does:
13:44:44 <warlord> touch Makefile || my_die "touch Makefile failed."
13:44:44 <warlord> make -f Makefile.am doxygen.cfg top_srcdir=$top_srcdir VERSION="$VERSION" >/
13:44:44 <warlord> dev/null 2>&1 || my_die "could not build doxygen.cfg."
13:45:26 <warlord> I.e., I dont actually run configure, I just use the Makefile.am rule to build doxygen.cfg
13:48:20 <jralls> So you need a standalone script to generate the doxygen.cfg. That shouldn't be a problem.
13:49:42 <warlord> that would work, too -- but using Makefile.am is how it's been done. I suppose I could do another branch: if [ -f Makefile.am ] ....
13:50:57 <jralls> Doesn't need a branch. If there's no configure there's no Makefile.am.
13:51:39 <luc14n0> jralls: it's alright. Now, yesterday I discovered why cmake was failing in open build service and not in travis ci (the problem I was having in that 'problem when linking libraries' bug). Our openSUSE cmake macros use the -Wl,--no-undefined flag and the build fails in libgncmod-gnome-utils.so. So apparently there is some library that is being used but not being linked to. What do you think?
13:52:59 *** wol1 has quit IRC
13:53:56 <jralls> luc14n0: Possible but seems unlikely. What was the undefined symbol?
13:54:42 <warlord> jralls: except the commands are in different sections of the file .
13:57:05 <luc14n0> jralls: In about a couple hours I have a keyboard to reach for and I shall tell you.
13:57:33 <luc14n0> s/have/will have
13:59:09 <jralls> warlord: HAVE_CONFIGURE=no; if [ - f configure ]; then HAVE_CONFIGURE=yes; fi; if [ "x$HAVE_CONFIGURE" = "xyes"]; then ...; else ...; fi
13:59:42 <warlord> jralls: thjat's still multiple branches ;)
14:00:46 <jralls> warlord: OK, so decompose it into functions and branch once calling the appropriate functions.
14:01:10 <warlord> too much effort.
14:01:40 <jralls> Well that's no you then. ;-)
14:02:21 <jralls> As for the doxygen.cfg target in libgnucash/doc/Makefile.am, it has two lines:
14:02:54 <jralls> (Three, sire)
14:03:09 <jralls> @echo "doc: " $$(pwd) && echo "distdir: " $$(echo $(distdir))
14:03:09 <jralls> rm -rf html refman.pdf
14:03:09 <jralls> $(DOXYGEN) doxygen.cfg
14:04:41 <jralls> Oh, wait, that's wrong. that's the doc target that depends on doxygen.cfg. The doxygen.cfg target has four lines (five is right out):
14:04:57 <jralls> rm -f $@.tmp
14:04:58 <jralls> sed < $< > $@.tmp \
14:04:58 <jralls> -e 's:[@]top_srcdir[@]:${top_srcdir}:g; s:[@]VERSION[@]:${VERSION}:g'
14:04:59 <jralls> mv $@.tmp $@
14:05:42 *** puck has quit IRC
14:06:07 <jralls> Maybe you should just put that in the script substituting "doxygen.cfg" as appropriate.
14:06:16 <gjanssens> fell: RL accounting is done for now, but RL is still keeping me pretty busy...
14:07:27 <fell> gjanssens saw my comment on PR 265?
14:10:17 <warlord> jralls: I could probably do that.
14:11:14 *** fabior has quit IRC
14:12:39 *** puck has joined #gnucash
14:16:17 <jralls> warlord: Can you tell when the doxygen build last ran?
14:16:27 <jralls> On master, that is.
14:17:33 *** frakturfreak has joined #gnucash
14:18:03 <gjanssens> fell: I answerred partly on the PR comment just now.
14:18:15 <fell> Thx
14:18:58 <gjanssens> As to which strategy to follow there are several aspects
14:19:11 <gjanssens> 1. we should separate maint and unstable translations.
14:20:01 <gjanssens> I would discourage further translation on maint, but if translators insist for the last maint release we should inform them clearly that work will not be picked up automatically by unstable any more.
14:20:09 <gjanssens> So it would have to be redone for 2.7/3.0
14:21:05 <gjanssens> 2. If we accept translations on maint, we probably have to ignore them in the merge to unstable.
14:21:50 <gjanssens> I would decide on a case by case basis. But if like in the ru.po case there were more than 600 unresolved merge conflicts, I'd consider it a lost cause and just keep what's on unstable.
14:22:22 *** harshitaneja has quit IRC
14:22:32 <warlord> today ;)
14:22:50 <gjanssens> 3. As for the pre-merge of goffice, this mostly affects translators that already started translating from the 2.7.3 gnucash.pot file.
14:23:02 <jralls> warlord: OK, how about "successfully".
14:23:55 <gjanssens> I have merged the existing goffice translations into the po files for 2.7.4, so all translators that will start from there would see these imports.
14:25:14 <gjanssens> 4. If we get more PR's from translations that started from before the goffice import, these files may have merge conflicts
14:25:20 <fell> I don't know if TP only offers our pot or also the updated po's.
14:25:52 <gjanssens> TP points to the tarball that goes with the gnucash.pot file. That should contain the right po files.
14:26:21 <fell> I remember after changing from ISO to UTF, I had do convert some files later again.
14:27:02 <gjanssens> The merge conflicts should be handled as follows: all merge conflicts in goffice strings should be resolved in favour of the goffice translations, and we should inform the translator of this. All other merge conflicts should be resolved in favor of the PR being merged.
14:28:10 <fell> But only for not (empty or fuzzy) goffice msgstr's
14:28:26 <gjanssens> fell: I can imagine the translators don't always start from the most recent po file but rather merge gnucash.pot with their local po file. Should we encourage them to use the one from the tar ball ?
14:28:46 <gjanssens> And yes, you're right about the non-empty/fuzzy msgstrs
14:29:51 <fell> Yep, at least should they "merge" them if they changed theirs.
14:30:00 <jralls> Each release contains the latest po's from the TP and a newly-generated gnucash.pot that is passed to the translators to merge and update. Few actually do, most of the TP translations are quite stale.
14:30:41 <jralls> But even the most up-to-date in our repo reflect the previous release's gnucash.pot.
14:31:16 <fell> Perhaps we should sende a explicit message to gnucash-user bccing the list of last translators?
14:33:11 <gjanssens> Yes, translations will always run behind a bit, ideally only from the previous release
14:34:23 <jralls> I told Benno about 2.7.3 and so far the Spanish team has mad a translation for it. I'll have to be careful not to put that in 2.6.20.
14:35:42 <gjanssens> Sending a explicit call for translations on user and devel wouldn't hurt. I'd wait until 2.7.4 is out though so the goffice po merge would be in the tarball the translators get
14:37:01 <gjanssens> jralls: I haven't looked at the Spanish translation yet. Did they manually translate the goffice strings as well ? If not, you'll have to take care not to overwrite the translations that got imported from goffice.
14:39:56 <jralls> gjanssens: No, the TP works exclusively from gnucash.pot in a release tarball. They won't go to git. Probably the best thing to do would be to make feature branch from before your potfile merge, copy in the TP translation, and then merge that into unstable so that git identifies the conflicts.
14:41:23 <jralls> fell: Did you merge your mnemonic updates?
14:42:22 <fell> The not string breaking part should be there
14:44:15 <fell> There are both improved: https://translationproject.org/PO-files/es/gnucash-2.7.3.es.po and https://translationproject.org/PO-files/es/gnucash-2.6.19.es.po
14:45:04 <fell> 2.7.3 has 5194 / 5194
14:45:46 <jralls> Ah, I didn't realize those were links. That makes it easy.
14:49:17 <gjanssens> jralls: I think that would be the safest approach indeed.
14:49:22 <jralls> fell: What if you go ahead and push your string-breaking changes now? That way I can fix both the goffice and mnemonic changes in one go and tell the Spanish translator about it, and all will be good for 2.7.4.
14:49:57 <fell> Mark Haanen (nl) completed 2.6.19 but not 2.73
14:50:25 <gjanssens> jralls: I was just about to propose I try and merge the Spanish translation, but I'm missing out on the "string-breaking" changes
14:50:27 <gjanssens> What is that ?
14:52:26 <jralls> gjanssens: fell has some mnemonic fixes in his local repo.
14:52:56 <fell> No, they are curently only comments in de.po
14:53:07 <fell> in stable
14:54:45 <jralls> Oh, I thought that you had them ready on a feature branch. And if they're in stable then they might be stale as well. Hmm.
14:55:38 <fell> Too many changes for me at the same time: dir structure, build tools, ...
14:55:39 <jralls> fell: How long do you think it would take to check them in unstable and turn them into a commit?
14:56:43 <fell> Not until this weekend
14:56:45 <jralls> Yeah, speaking of the dir change I just confirmed that it broke the doxygen build.
14:58:01 <gjanssens> jralls: interesting, what's the error ?
14:58:10 <jralls> OK. Maybe we should slip 2.7.4 until next week.
14:58:44 <gjanssens> Or more quickly: what's the command to build doxygen ? ninja-build doxygen ?
14:58:46 <jralls> make[1]: *** src/doc: No such file or directory. Stop.
14:59:54 <jralls> The target in the makefile is ${DOXYGEN} doxygen.cfg.
15:01:24 <fell> make doc or ninja doc
15:01:37 <fell> says our faq
15:02:56 <jralls> fell: Which is not currently correct because CMake doesn't have a doc target.
15:03:37 * gjanssens just discovered that as well
15:04:04 <fell> So, whoever inserted ninja there, did'nt test it ;-)
15:04:21 <jralls> However, with the appropriate substitutions into doxygen.cfg.in it does work when run by hand.
15:04:51 <gjanssens> So it's not only the dir change that broke the build but removing the autotools
15:05:24 <gjanssens> jralls: and with the appropriate substitutions the doxygen build works ? Or do you still get errors regarding src/doc ?
15:05:33 <gjanssens> (which should now be libgnucash/doc)
15:06:10 <gjanssens> A quick grep suggests src/doc is still used in several spots though, but most of them look like plain text rather than commands
15:07:12 <jralls> It works, as I just said. doxygen.cfg.in has been modified to use libgnucash instead of src. However, it only looks at libgnucash.
15:08:08 <gjanssens> Oh, right. So any doxygen info in gnucash or common is missing ?
15:08:23 <jralls> Yup. Or bindings.
15:08:34 <gjanssens> Those too...
15:09:40 <jralls> I think the references to src/doc are comments referring to the obsolete documentation. We need to make sure that all of that that's still relevant is reflected in the right places in doxygen comments and then nuke libgnucash/doc.
15:11:58 <gjanssens> jralls: to get the other paths in doxygen you'll have to add them in doxygen.cfg.in around line 655
15:12:00 *** oozer has quit IRC
15:12:18 <gjanssens> It used to read INPUT = <topsrcdir>/src
15:12:25 <gjanssens> That would include all subdirectories of src
15:12:26 <jralls> Yeah, I know. I just did that and ran doxygen. No errors, checking the generated html.
15:12:46 <gjanssens> Heh, I'll go back to the Spanish translation then :)
15:15:27 <jralls> Looks like it mostly worked. Good, it won't take long to move it to CMakeLists.txt.
15:16:56 <jralls> So back to translations. Should we slip 2.7.4 a week to give fell time to get his mnemonic changes in?
15:18:43 <fell> jralls: if you open the preferences, can you enter umlauts in account separator?
15:19:23 <jralls> fell: On what platform?
15:19:31 <fell> MacOs
15:20:06 <jralls> fell: No, on the mac keyboard the umlaut is a combining character. It can't be entered stand-alone.
15:20:39 <fell> Because the page uses Mnenomics
15:21:12 <jralls> Although I can make umlaut-space, ¨ which looks like a stand-alone umlaut.
15:22:45 <jralls> Does it use _u as a mnemonic?
15:23:12 <fell> No, I was talking about the <alt>+char
15:23:23 <fell> composition
15:24:34 <fell> We use "_e" and "_a" in the currency section
15:25:27 <fell> So what happens, if you try to enter alt+e or a as character separator?
15:26:28 <jralls> I get an accent-grave for the first and å for the second.
15:27:18 <fell> and it is not jumping to local/choose currency?
15:28:42 <jralls> Nope. And none of the mnemonics are underlined, either. Hmm, should I have tested in 2.7 or 2.6? I did in 2.6.
15:29:06 <fell> yes, I am in 2,6
15:29:52 <jralls> So it looks like it's just ignoring the mnemonics. That's Gtk2 though, let me try 2.7 and see what happens.
15:29:53 <fell> Loc_ale and Ch_oose are our labels
15:30:52 <jralls> On 2.7 it jumps when I try to enter å.
15:31:39 <fell> Could there be some switch to change the behavior?
15:33:11 <jralls> I think Gtk has a switch to disable mnemonics entirely. I can also look into patching it to use <ctrl><alt> on macs.
15:34:55 <jralls> I see Mitch Natterer seems to be present on #gimp. He's the author of GtkModifierIntent, I'll pop over there and ask him if he's thought about it.
15:46:53 *** luc14n0 has quit IRC
15:46:58 *** luc14n0 has joined #gnucash
15:50:04 *** luc14n0 has quit IRC
15:50:09 *** luc14n0 has joined #gnucash
15:58:31 <gjanssens> The Spanish translation turned up an even greater number of unresolved conflicts: over 3000....
15:59:57 <fell> xaviers new 2.7.3?
16:00:12 <gjanssens> yes
16:00:21 <fell> or his 2.6.19
16:01:11 <gjanssens> Oh, no the attempt to merge maint into unstable after applying his 2.6.19 to maint first
16:01:40 <gjanssens> But I have followed the same conflict resolution as described above.
16:02:41 <gjanssens> There were 26 differences between the goffice auto import and his goffice translations
16:03:20 <jralls> Oh, 2.7.3 already has the goffice translations? Then we're good, we can just use it.
16:03:36 <gjanssens> I will mail him to review these again and if so correct them for the next merge
16:04:18 <gjanssens> jralls: not quite. 2.7.6 has the goffice strings, I have merged the translations after 2.7.3 got released
16:04:31 <gjanssens> 2.7.3*
16:04:36 <gjanssens> not 2.7.6
16:05:15 <gjanssens> But translators working from the 2.7.3 pot file have been independently translating the goffice strings and sometimes their translations differ slightly from the one I imported from goffice
16:05:33 <jralls> Right. We should go with theirs.
16:05:43 <gjanssens> In these cases I have always preferred the goffice translations
16:05:50 <gjanssens> Is that what you meant ?
16:06:20 <gjanssens> And I inform the translator of the inconsistencies so they can reverify
16:07:30 <jralls> Yes. If the translators have already translated the strings then I think we should prefer their translations. Your work could be viewed as a convenience for the ones who haven't yet translated them and a help to users for the translations that have fallen by the wayside.
16:07:55 <jralls> Which unfortunately is most of the TP ones.
16:07:55 <gjanssens> So I've been doing it the wrong way around :(
16:08:22 <jralls> Not really. There's only one translation of 2.7.3 that's been done.
16:08:23 <gjanssens> My reasoning was the goffice translations should be more mature, so their strings are likely more stable.
16:09:18 <gjanssens> But I can do it both ways. I can prefer our translator's strings but still inform them of the differences with goffice translations I found
16:09:30 <fell> we got already de, ru, es
16:09:52 <jralls> You certainly can.
16:10:02 <gjanssens> So the translator can optionally improve on his/her work for gnucash or decide the goffice translation is wrong
16:11:26 <jralls> Right. Though it might be more tactful to say "reconsider" instead of "improve".
16:12:02 * jralls wonders if Mechtilde might be listening in...
16:12:43 <gjanssens> :) That was not the way to bring it across to the translators. I would talk about comparing and deciding which translation would be best.
16:12:57 <gjanssens> Or preferred.
16:14:25 <gjanssens> The German translation was my first experiment. I did send Mechtilde a list of differences between her translation and the goffice one.
16:14:54 <jralls> And did she respond?
16:15:23 <Mechtilde> not yet, sorry
16:15:33 <Mechtilde> gjanssens, it was ok
16:16:14 <Mechtilde> I will look through it next time I work on gnucash translation
16:16:45 <gjanssens> Mechtilde: that's fine
16:16:50 <Mechtilde> does someone of you come to FOSDEM Brussels
16:16:52 <Mechtilde> ?
16:17:24 <gjanssens> I'm not sure yet. I have a training course the next week and it may be a bit too much at once.
16:17:42 <gjanssens> But usually I try to go there.
16:18:01 <gjanssens> In US notions of distance it's right in my back yard after all...
16:18:06 <jralls> One of these years perhaps I can make time for that. Not this one, unfortunately.
16:18:31 <gjanssens> jralls: that would be a great event
16:18:32 <jralls> I'd think that'd be in your back yard in anyone's notion of distance.
16:18:45 <gjanssens> :D
16:19:04 <gjanssens> If you can come one day, be sure to inform us in advance
16:19:25 <gjanssens> It may be a good opportunity to gather with a few gnucash devs on this side of the big pond
16:19:28 <Mechtilde> I will there at Apache Booth
16:19:52 <Mechtilde> most of the time
16:19:57 <gjanssens> Ok, I'll try to drop by (if I make it to fosdem)
16:22:03 *** fekepp has joined #gnucash
16:28:42 <fell> I am this year to busy with other things.
16:29:02 *** luc14n0 has quit IRC
16:29:07 *** luc14n0 has joined #gnucash
16:31:00 <fell> gjanssens, and the dutch po?
16:32:56 <fell> You should talk with Mark about the branches and might also look in gnucash-nl. I am not sure about betalingsvoorstel.
16:39:48 *** Mechtilde has quit IRC
16:53:10 *** meb has joined #gnucash
16:58:10 *** meb has quit IRC
17:09:31 <luc14n0> jralls: I'm using -Wl,-z,defs and here's what I got: FAILED: lib64/gnucash/libgncmod-gnome-utils.so
17:09:31 <luc14n0> ...
17:09:31 <luc14n0> gnucash/gnome-utils/CMakeFiles/gncmod-gnome-utils.dir/dialog-utils.c.o: In function `gnc_builder_connect_full_func':
17:09:31 <luc14n0> /home/abuild/rpmbuild/BUILD/gnucash-2.7.3/build/../gnucash/gnome-utils/dialog-utils.c:553: undefined reference to `dlsym'
17:12:46 <jralls> luc14n0: man 3 dlsym says that one needs to link with libdl.so, which should be a pretty basic part of your OS.
17:13:42 <gjanssens> jralls: your most recent commit adds Makefile.am to the dist list, but you didn't add that file in itself. Is is required for doxygen to work or did you intend to add something else ?
17:15:58 <jralls> I didn't intend to add it at all. I didn't even mean to change that line.
17:16:40 <gjanssens> Live code review :)
17:16:50 *** meb has joined #gnucash
17:17:53 <jralls> gjanssens: Thanks, fixed.
17:18:20 <gjanssens> You're welcome
17:20:06 *** pilotauto has joined #gnucash
17:20:32 <jralls> Almost pair programming, except that that works *before* the commit. ;-)
17:21:48 <luc14n0> jralls: hm hm, autotools used to pass the -ldl/ldld flag(s) to the linker
17:22:38 <gjanssens> jralls: that would be kind of hard when I don't see what you are typing until after the commit :s
17:23:02 <gjanssens> fell: What did you mean with betalingsvoorstel on gnucash-nl ?
17:23:46 <fell> his last posting on the dutch mailing list
17:24:36 <jralls> gjanssens: "Pair programming" usually means two programmers side-by-side. One codes, the other watches for mistakes.
17:25:15 <gjanssens> Ooh. I didn't know that concept.
17:26:07 <jralls> luc14no: So you need get the OpenSuSE folks to fiddle their CMake installation so that it does the right thing to. None of the other distros have reported that particular problem.
17:26:07 <gjanssens> fell: I believe Mark is correct. There's no support for online banking via gnucash in the Netherlands or Belgium. Unfortunately.
17:26:49 <jralls> gjanssens: It's part of https://en.wikipedia.org/wiki/Agile_software_development.
17:27:52 <gjanssens> jralls: interesting...
17:31:24 <fell> jralls OBS ass more targets than only *SuSE
17:31:43 <jralls> OBS?
17:32:14 <gjanssens> fell: It looks like Mark hasn't posted any translation work yet on the unstable branch. The po file on the TP is from the release date
17:32:25 <gjanssens> OBS = OpenSuse build service
17:33:27 <gjanssens> A build platform that allows to build and package for multiple platforms.
17:33:38 <fell> but an complete 2.6.19
17:33:57 <gjanssens> jralls: with link: https://build.opensuse.org/
17:34:24 <gjanssens> fell: yes I have just pushed a commit with that translation to the maint branch (and merged into unstable afterwards)
17:34:33 <fell> thx
17:35:02 <gjanssens> Or the code: http://openbuildservice.org/
17:35:16 <gjanssens> Time to go to bed here...
17:35:37 <jralls> Thanks for the ptr to OBS. Goodnight.
17:35:45 <gjanssens> It was nice chatting and collaborating again.
17:35:51 <gjanssens> See you
17:36:30 *** gjanssens has quit IRC
17:38:10 <jralls> fell: Here's luc14n0's build: https://build.opensuse.org/package/show/home:luc14n0:branches:GNOME:Next/gnucash. Looks to me like it's OpenSuSE and that it's failing only on 32-bit builds.
17:39:13 <luc14n0> I'm using this one to make test: https://build.opensuse.org/package/show/home:luc14n0:branches:home:luc14n0:branches:GNOME:Next/gnucash
17:40:05 <luc14n0> if I don't use -Wl,--no-undefined it build ok, but something would still be missing
17:41:14 <luc14n0> autotool's configure used to : AC_CHECK_FUNCS(dlsym,,[
17:41:15 <luc14n0> AC_CHECK_LIB(dl, dlsym, DL_LIB="-ldl",[
17:41:15 <luc14n0> AC_CHECK_LIB(dld, shl_load, DL_LIB="-ldld",[
17:41:15 <luc14n0> AC_CHECK_FUNCS(dlsym, DL_LIB="")
17:42:47 <jralls> luc14n0: I don't see that in our configure.ac. Where did you find it?
17:43:45 <luc14n0> that's the 2.7.3 github's release tarball
17:45:09 *** oozer has joined #gnucash
17:46:47 <jralls> OK, I found it with a better grep.
17:49:29 <luc14n0> I'm not that used to cmake but I think it uses CHECK_LIBRARY_EXISTS in situations like this.
17:53:37 *** frakturfreak has quit IRC
18:02:55 <jralls> Hmm, so we're not even checking. It's actually conditionally compiled but the condition is unconditionally set for Unix.
18:04:06 <jralls> And while -ldl doesn't show up on the link line it's there in ldd output so it's presumably being pulled in by something else, probably glib or gtk.
18:05:31 <jralls> luc14n0: If you build without -Wl,--no-undefined does libdl show up in `ldd lib/gnucash/libgncmod-gnome-utils.so | grep libdl.so` ?
18:05:56 <luc14n0> Let me do the test.
18:19:03 <jralls> luc14n0: If set that flag in CMAKE_C_FLAGS on Debian 9 it errors out with a bunch of symbols linking libgncmod-app-utils.so, long before it gets to libgncmod-gnome-utils.so.
18:21:19 <fell> jralls, I wonder if anybody needs GNOME2_STATUS, referenced in Changelog 2004-6, linking to bugzilla &version=SVN
18:23:51 <jralls> fell: Not any more. If you follow the link you get "Zarro Boogs found."
18:24:24 <fell> because we replaces svn by git*
18:27:40 <fell> with git-maint I get 17, but see no relation to gnome2-port
18:30:02 <jralls> Huh? The link points to bugzilla.
18:31:03 <jralls> Oh, I see what you mean.
18:31:17 <fell> and in bugzilla we replaced version SVN by git-maint and git-master
18:33:35 <jralls> Right. So the search is basically everything for version SVN with no other qualifiers. That's useless.
18:38:26 <luc14n0> jralls: yes, it does show up. Excuse my ignorance but -ldl goes in CMAKE_C_FLAGS instead of CMAKE_SHARED_LINKER_FLAGS ?
18:42:13 <jralls> luc14n0: I'd go with LINKER_FLAGS if you're going to add it to the command line.
18:42:54 <luc14n0> oh, I get it!
18:45:01 <luc14n0> Let me test what happens if I just add -ldl as well. And I'm going to setup a travis ci build with ninja in verbose mode.
18:45:05 <jralls> luc14n0: OTOH if you're going to submit a patch you should use find_library() and add it to the gnome-utils target in gnucash/gnome-utils/CMakeLists.txt.
18:45:45 <luc14n0> jralls: yeah, I saw this option too in my search.
18:47:17 <luc14n0> After dinner I will start to work on this.
18:48:00 <luc14n0> jralls: By the way, you guys will move issue tracking to github?
18:48:58 <jralls> Probably not. github's tracker doesn't work very well for large bug sets. We'll probably have to migrate to a self-hosted BZ when Gnome gets close to shutting theirs down.
19:08:22 <luc14n0> jralls: interesting, if I append -ldl, no problems
19:12:45 *** oozer_ has joined #gnucash
19:12:46 *** oozer has quit IRC
19:13:13 *** chris has joined #gnucash
19:15:26 <chris> .
19:15:26 <gncbot> chris: Sent 9 hours and 17 minutes ago: <warlord> I don't think it ALWAYS did that, unfortunately. I think that boolean was added later. We could attempt to build a filter that would add that flag on file conversion from an older version going forward?
19:19:00 <jralls> lic14n0: No surprise, eh?
19:22:27 <luc14n0> This patch should be easy then. No it's time to have dinner.
19:23:33 *** chris has quit IRC
19:43:11 *** oozer_ has quit IRC
19:54:08 *** kus has joined #gnucash
20:11:35 *** pilotauto has quit IRC
20:13:12 *** pilotauto has joined #gnucash
20:16:23 *** pilotauto has quit IRC
21:18:47 *** kus has quit IRC
21:49:54 *** AndroUser has joined #gnucash
22:04:38 *** AndroUser has quit IRC
22:30:35 *** ArtGravity has quit IRC
22:31:45 *** kael has joined #gnucash
22:42:14 *** Aussie_matt has joined #gnucash
22:51:25 *** Aussie_matt has quit IRC
23:06:12 *** kael has quit IRC
23:39:15 *** Cuare has quit IRC