2018-09-09 GnuCash IRC logs
01:03:45 *** Mechtilde_ has quit IRC
01:13:28 *** Mechtilde_ has joined #gnucash
01:58:49 *** Aussie_matt has quit IRC
02:26:02 *** fell has joined #gnucash
03:21:13 *** storyjesse has joined #gnucash
03:24:20 *** jotrago1 has joined #gnucash
03:24:43 *** jotrago has quit IRC
03:24:44 *** jotrago1 is now known as jotrago
03:42:24 *** boldstripe has joined #gnucash
03:54:50 *** boldstripe has quit IRC
03:56:25 *** Aussie_matt has joined #gnucash
03:57:32 *** boldstripe has joined #gnucash
03:57:45 *** finster has joined #gnucash
04:02:05 *** boldstripe has quit IRC
04:30:55 *** jotrago has quit IRC
05:19:22 *** gjanssens has joined #gnucash
05:19:22 *** ChanServ sets mode: +o gjanssens
05:26:19 *** bertbob has quit IRC
05:27:00 *** ncv__ has joined #gnucash
05:29:40 *** bertbob has joined #gnucash
05:38:45 *** Mechtilde_ has quit IRC
05:40:40 *** storyjesse has quit IRC
05:42:17 *** storyjesse has joined #gnucash
05:45:28 *** Mechtilde_ has joined #gnucash
05:54:23 *** bertbob has quit IRC
05:56:44 *** bertbob has joined #gnucash
06:10:28 *** kritzefitz has joined #gnucash
06:16:13 *** bertbob has quit IRC
06:21:41 *** oozer has joined #gnucash
06:24:32 *** bertbob has joined #gnucash
06:27:38 *** bertbob has quit IRC
06:42:22 *** bertbob has joined #gnucash
06:51:12 *** jotrago has joined #gnucash
07:43:44 *** oozer has quit IRC
08:22:46 <fell> While playing with Calibre, I found https://books.google.de/books?id=fMktswEACAAJ&dq=gnucash&hl=de&sa=X&ved=0ahUKEwjn4JXg863dAhVolIsKHTxsDP0Q6AEIPDAC
08:56:31 *** oozer has joined #gnucash
09:06:05 *** Aussie_matt has quit IRC
09:12:03 *** O01eg has joined #gnucash
09:24:32 *** ncv_ has joined #gnucash
09:25:44 *** ncv__ has quit IRC
09:30:04 *** ncv__ has joined #gnucash
09:31:13 *** ncv_ has quit IRC
09:48:58 *** storyjesse has quit IRC
09:59:48 *** jotrago1 has joined #gnucash
10:00:12 *** jotrago has quit IRC
10:00:12 *** jotrago1 is now known as jotrago
10:17:13 <gjanssens> fell: sweet :)
10:19:21 <chris> jralls #398 and #406 are now independent. #406 creates the tests, #398 performs the changes that still pass the tests.
10:29:26 * chris should be studying instead of hacking
10:40:15 *** User has joined #gnucash
10:44:49 *** User has quit IRC
11:37:31 *** jotrago1 has joined #gnucash
11:37:43 *** jotrago has quit IRC
11:37:43 *** jotrago1 is now known as jotrago
11:39:52 *** fell has quit IRC
11:41:35 *** JayC has quit IRC
11:43:29 *** JayC has joined #gnucash
11:50:48 *** fell has joined #gnucash
12:02:53 *** jotrago1 has joined #gnucash
12:03:17 *** jotrago has quit IRC
12:03:17 *** jotrago1 is now known as jotrago
12:05:53 *** gncbot sets mode: +o fell
12:07:45 *** Mechtilde_ has quit IRC
12:13:52 *** Mechtilde_ has joined #gnucash
12:27:54 *** ncv__ has quit IRC
12:31:32 *** jotrago1 has joined #gnucash
12:31:51 *** jotrago has quit IRC
12:31:51 *** jotrago1 is now known as jotrago
12:54:53 *** finster has quit IRC
13:54:30 *** warlord has quit IRC
14:16:47 <jralls> gjanssens: I noticed a problem with my quickfill selection fix: When entering an account, if you do an account separator the selection is adjusted to after the separator. So far so good. But if you then press <shift> to capitalize the next account name it gets put back to wherever it was after the last character typed in the quickfill stage, not so good.
14:17:09 <jralls> gjanssens: Do you know where the selection gets adjusted after typing the account separator?
14:17:52 <gjanssens> jralls: let me check
14:22:41 *** User has joined #gnucash
14:25:44 *** User has quit IRC
14:25:48 <gjanssens> jralls: This should start somewhere in gnc_combo_cell_direct_update()
14:26:22 <gjanssens> gnucash/register/register-gnome/combocell-gnome.c
14:28:16 <gjanssens> It checks if the account separator is typed (which was stored in box->complete_char earlier on) and then does a number of checks and updates the quickfill stuff
14:29:32 <jralls> Ah, box->complete_char is the piece I was missing: I was looking for account_separator.
14:31:54 <jralls> Thanks.
14:32:50 <gjanssens> Np
14:37:46 *** frakturfreak has joined #gnucash
15:25:38 *** Mechtilde_ has quit IRC
15:59:44 *** warlord has joined #gnucash
16:05:10 <warlord> .
16:49:58 *** frakturfreak has quit IRC
16:58:21 <gjanssens> jralls: considering you probably have fixed the most critical usability issues with our register and input methods, I think I will defer further work on it myself.
16:58:47 <gjanssens> It's too much of a hairy mess to carefully refactor in C.
16:59:06 <jralls> gjanssens: OK. I hope you're right about fixed and I agree it's a bit of a mess.
16:59:38 <gjanssens> I intend to revisit it for 4.0 at which point I plan to redo the data model.
17:00:42 <gjanssens> Yesterday and today I spent some time hunting memory leaks.
17:01:04 <gjanssens> That was an interesting challenge on its own. Certainly in the C world...
17:01:07 <jralls> With Valgrind, clang, or by hand?
17:01:17 <gjanssens> With heaptrack
17:01:49 <gjanssens> https://www.kdab.com/heaptrack-v1-0-0-release/
17:02:08 <gjanssens> I'm impressed with this tool.
17:02:23 <gjanssens> Especially the flame graph is very handy.
17:02:46 <gjanssens> I did fix a few issues already but there are several more.
17:03:31 <gjanssens> The catch with memory leaks is of course the tool can only detect them if memory is allocated, so when parts of the code are actually used.
17:03:35 <jralls> I bet there are a *lot* more.
17:03:49 <gjanssens> Yeah indeed.
17:04:03 <gjanssens> I meant to say there are several more you'll see by only starting and stopping gnucash...
17:04:31 <gjanssens> And then I saw even more by simply opening and closing a register (the worst offender there should be fixed now)
17:05:27 <jralls> And then there's QofQuery. I saw how easy it is to leak with that while reviewing Bob's code yesterday.
17:06:00 <jralls> Plus how inefficient. Lots of deep copies where smart pointers would be much cheaper.
17:11:44 <gjanssens> Yeah, speaking of inefficiency: I'm experimenting with a huge book (50000 transactions, 170000 splits)
17:11:57 <gjanssens> Opening an xml version is reasonable: 1m20s
17:12:19 *** kritzefitz has quit IRC
17:12:29 <gjanssens> Opening an sqlite version is horrible: I'm waiting more than 15 mins already
17:12:48 <gjanssens> The user that provided me with this book says it takes over an hour to load.
17:13:14 <gjanssens> And that's starting from an SSD drive and plenty of free RAM...
17:14:39 <jralls> Hmm, that's a bit bigger than my primary file (34821 txns and 71038 splits).
17:15:50 <jralls> How many slots does it have? That's the real performance killer.
17:16:45 <gjanssens> Roughly 265000
17:17:15 <jralls> OK, no surprise then. I've only got 41452.
17:17:35 <gjanssens> Yeah that does make a difference.
17:18:02 <gjanssens> It's a book with lots of business objects, which I believe you don't use
17:19:10 <jralls> Also true. But those are pretty straightforward tables IIRC. They should load pretty fast.
17:20:39 <jralls> If you run with --log gnc.backend.dbi=debug it will put the queries in the tracefile with timestamps and you can pretty easily see which ones are taking the most time.
17:21:30 <jralls> I used that as an ad-hoc profiler last spring when I was working on speeding up SQL loading on Windows.
17:34:01 <gjanssens> Ok, I think I'll dig into that one of the next days.
17:34:59 <gjanssens> BTW we now have test-report-utilities failing
17:35:10 <gjanssens> I don't know if this is from your merges of from my code.
17:35:23 <gjanssens> I'm too tired now to look into it...
17:35:32 <gjanssens> I'm off to bed instead :)
17:35:51 <jralls> OK. That's from merging Chris's code. Shall I fix your oops that was reported on gnucash-devel as well?
17:38:06 <gjanssens> Just did that myself
17:38:16 <gjanssens> Oops indeed...
17:38:18 <gjanssens> Night!
17:38:26 <jralls> Goodnight!
17:39:45 *** gjanssens has quit IRC
18:31:31 *** pilotauto has joined #gnucash
19:18:54 *** fell has quit IRC
19:30:52 *** fell has joined #gnucash
20:06:48 *** gncbot sets mode: +o fell
20:45:54 *** fell has quit IRC
21:36:43 *** storyjesse has joined #gnucash
21:40:33 *** oozer has quit IRC