2024-01-14 GnuCash IRC logs
01:08:11 *** fell has quit IRC
01:09:30 *** fell has joined #gnucash
01:09:31 *** ChanServ sets mode: +o fell
03:55:17 *** jony has joined #gnucash
03:57:09 *** jony has quit IRC
03:57:41 *** jony has joined #gnucash
04:01:42 *** jony has quit IRC
04:45:09 *** gjanssens has joined #gnucash
04:45:09 *** ChanServ sets mode: +o gjanssens
04:54:42 <sunyibo> https://bpa.st/SP5A can anyone make sense of this?
04:54:46 *** sunyibo has quit IRC
05:11:01 *** sunyibo has joined #gnucash
05:11:01 *** ChanServ sets mode: +v sunyibo
08:21:50 *** CWehli has joined #gnucash
10:11:56 <warlord> sunyibo, perhaps there is another sub-dependency that didn't get pulled in?
12:03:29 *** AdrienM has quit IRC
12:04:21 *** AdrienM has joined #gnucash
12:04:21 *** ChanServ sets mode: +v AdrienM
12:09:15 <jralls> sunyibo: You must have two perls installed and GnuCash is finding the one that isn't in the gnc-fq-update shebang.
12:16:05 <jralls> sunyibo: The shebang perl is configured at build time and GnuCash looks along the path, so if you're using the distro-packaged GnuCash the former is probably /usr/bin/perl. If that's not the first one encountered by traversing $PATH you'll get the indicated error.
12:35:15 *** AdrienM has quit IRC
12:35:22 *** AdrienM has joined #gnucash
12:35:23 *** ChanServ sets mode: +v AdrienM
13:51:33 *** NoobAlice has joined #gnucash
14:04:38 *** CWehli has quit IRC
14:11:30 <sunyibo> ok, thanks. And how can I fix that?
14:23:26 <jralls> sunyibo, either delete the extra perl installation or rearrange your environment so that you have to explicitly invoke it by path.
14:24:27 <jralls> Since you had to ask the first is probably the better choice, and in future install software solely with the package manager.
14:25:44 <sunyibo> thanks. I don't recall doing otherwise
14:28:06 <jralls> Hmm. In a command shell run "head -n1 /usr/bin/gnc-fq-update" and "which perl" (without the quotes). Do they print the same path?
14:31:13 <sunyibo> https://bpa.st/QNMA
14:32:29 <sunyibo> I would say so
14:37:25 <jralls> You'd be correct. Try 'perl -MJSON::Parse -e""'. Keep the double-quotes, leave out the single-quotes. It should just return.
14:38:49 <sunyibo> https://bpa.st/IS6A
14:45:03 <jralls> That's probably good enough, but to make sure it's actually trying to load the module try `perl -MJSON::Parse -e'print "\n";'` This time both the single and double quotes stay, it's the back-ticks that get left off.
14:46:18 <sunyibo> https://bpa.st/SUGA
14:50:04 <sunyibo> weird https://bpa.st/AEGQ
14:51:31 <jralls> Try the perl -MJSON... command again.
14:52:26 <sunyibo> https://bpa.st/5FDA
14:52:44 <sunyibo> and it works
14:52:47 <sunyibo> but why?
14:55:18 <jralls> Because you had an obsolete JSON::Parse in your home directory and that was in front of the system directories in the perl module path (perl global variable @INC).
14:56:06 <sunyibo> and how did it get there? gnc-fq-update?
14:57:11 <jralls> Maybe, from running it without sudo. What else is in /home/alex/perl5?
14:59:24 <sunyibo> https://bpa.st/5OBA . Also, may I humbly suggest, to "simply" integrate finance::quote into gnucash? Everytime I catch up with my accounting, I spend a considerable time to configure online quotes...
15:02:38 <jralls> OK, very likely from running gnc-fq-update without sudo. Don't do that, you really want everything in the system directories so cpan doesn't get confused.
15:02:54 *** palerider has joined #gnucash
15:03:02 <sunyibo> I couldn't find an occurence of running gnc-fq-update without sudo. But I recovered the home folder from an old archlinux install . Maybe that's the culprit
15:06:30 <jralls> That's very likely the culprit, because the JSON::Parse error indicated an older version of libperl. But the F::Q in your listing was the latest 1.59 that Bruce released only a couple of weeks ago.
15:09:19 <sunyibo> well, thanks a bunch anyhow!
15:32:09 *** palerider has left #gnucash
15:37:10 <jralls> Well, that's annoying. I finally got a build of webkit on Windows that doesn't crash as long as you run it in the debugger. Launch it from the start menu, open a chart, and bang.
16:14:02 <warlord> jralls, Uggh. What if you run it under Valgrind? I bet there is some use of uninitialized data that works fine in the debugger because it gets zero'd out, but in real life it points into lala land.
16:55:28 <jralls> warlord, does valgrind work under mingw32? I'm under the impression it doesn't.
16:58:34 <warlord> Apparently, yes: https://sourceforge.net/p/valgrind4win/wiki/DevelopmentEnvironment/
16:59:24 <warlord> It might need additional pieces.
16:59:49 <jralls> I'm skeptical, it was last updated in 2016, but I'll give it a try.
17:01:50 <warlord> In researching, it may or may not work.. Hard to tell.
17:02:11 <warlord> I did find this page with links to other tools, or the suggestion to run the app under wine under valgrind: https://stackoverflow.com/questions/413477/is-there-a-good-valgrind-substitute-for-windows
17:03:58 <jralls> Actually, no, I don't think I can even try. The additional pieces: MinGW classic from 2012, the Win7 SDK, .Net v4. And that SO is also 11 years old.
17:06:28 <warlord> Could we run Gnucash (or a test case) under wine?
17:09:23 <jralls> Would that get us anywhere?
17:09:50 <jralls> But I got it to crash in the debugger by launching from the Start menu and attaching the process.
17:10:37 <warlord> That does lead me to still believe it is an initialization error.
17:14:09 <jralls> Initialization of what?
17:17:13 <jralls> The debugger still doesn't have symbols, compiling with enable-debug or -g makes the binary too big to load, and enable-debug-symbols=full doesn't do anything without --enable-debug.
17:19:25 <jralls> Worse, the stack trace ends further down than before, so it's not even helpful.
19:31:06 <warlord> Weird!
20:00:03 *** raghukamath has quit IRC
22:07:58 *** raghukamath has joined #gnucash
22:07:58 *** ChanServ sets mode: +v raghukamath
22:32:56 *** NoobAlice has quit IRC
23:56:46 *** warlord has quit IRC
23:56:59 *** warlord has joined #gnucash
23:56:59 *** gncbot sets mode: +o warlord