2015-03-28 GnuCash IRC logs

00:03:33 *** GabrieleV_ has joined #gnucash
00:04:04 *** GabrieleV has quit IRC
00:04:05 *** GabrieleV_ is now known as GabrieleV
00:44:12 *** ErKa has quit IRC
01:03:33 *** GabrieleV_ has joined #gnucash
01:03:36 *** GabrieleV has quit IRC
01:03:37 *** GabrieleV_ is now known as GabrieleV
01:04:56 *** josePHPagoda has quit IRC
01:09:23 *** ArchangelXL has joined #gnucash
02:24:14 *** ArchangelXL has quit IRC
02:25:00 *** fell__ has quit IRC
02:25:24 *** fell__ has joined #gnucash
02:25:24 *** gncbot sets mode: +o fell__
02:46:51 *** StuM has quit IRC
03:03:07 *** GabrieleV_ has joined #gnucash
03:03:14 *** GabrieleV has quit IRC
03:03:14 *** GabrieleV_ is now known as GabrieleV
03:03:40 *** bozonius has joined #gnucash
03:04:27 <bozonius> posted an invoiced item to wrong income account. Any way to change this without damaging the data?
03:27:51 *** MechtiIde has joined #gnucash
03:43:51 *** len has quit IRC
04:09:39 *** GabrieleV_ has joined #gnucash
04:09:42 *** GabrieleV has quit IRC
04:09:42 *** GabrieleV_ is now known as GabrieleV
04:40:36 *** Krzysiek_K has joined #gnucash
05:45:31 *** fabior has joined #gnucash
05:53:08 *** gjanssens has joined #gnucash
05:53:09 *** gncbot sets mode: +o gjanssens
05:57:06 <gjanssens> fell__ I only saw your message regarding the locale wiki page today.
05:57:34 <gjanssens> I read your changes - they look fine to me. Thanks
06:10:09 * Simon notices that the database schema is finally not using TEXT for everything
06:12:41 *** rubdos has joined #gnucash
06:18:30 * Simon tries to figure out why xaccAccountTreeForEachTransaction is not run in a deterministic order
06:18:50 <Simon> almost every time I open and save the same xml file, two transactions swap around
06:19:33 <Simon> they're not even on the same day or for the same amount or have the same description, but they are each for the same account on one side of the single split
06:20:51 *** StuM has joined #gnucash
06:22:20 <Simon> it looks like they are the first two transactions written out for that account
06:27:09 *** misha_ has joined #gnucash
06:34:40 *** misha has quit IRC
06:36:21 <Simon> one of them has a memo of "" and the other has no memo
06:42:01 <Simon> these aren't even transactions... they're scheduled transactions
06:43:35 * Simon wonders just how much time gnucash spends calling g_utf8_collate("","")
06:45:16 <Simon> if I had to guess I'd say this code in http://s85.org/cZZrczoI is wrong because it operates differently depending on whether you call xaccSplitOrder(a,b) or xaccSplitOrder(b,a)
06:45:59 <Simon> if qof_book_use_split_action_for_num_field(xaccSplitGetBook(sa)) != qof_book_use_split_action_for_num_field(xaccSplitGetBook(sb)) then the sort ordering could be inconsistent
06:46:29 <Simon> although it refers to the book which should be the same
06:48:06 <Simon> is there a reason why xaccSplitOrder doesn't just use the guid?
07:09:03 <Simon> it only seems to call xaccTransOrder and xaccSplitOrder if I open the scheduled transaction, otherwise it never calls them
07:27:02 <Simon> gnc_account_insert_split is calling g_list_prepend and then the list is never sorted
07:27:30 <Simon> this is the only scheduled transaction account that has two scheduled transactions against it
07:29:43 *** fl0rent has joined #gnucash
07:35:56 <Simon> it looks like it just never calls xaccAccountSortSplits for those accounts unless it's displayed
07:39:36 <Simon> that gets called for the main list of accounts because they're all present in the tree that's always open
07:39:47 <Simon> but it won't call it for the scheduled transaction accounts unless you open the scheduled transaction editor
08:03:06 *** pacon has joined #gnucash
08:09:57 <Simon> guile should really store the source file for comparison, given how long it takes to compile it...
08:15:02 *** himaxx has joined #gnucash
08:21:53 *** himaxx has quit IRC
08:28:14 *** fl0rent has quit IRC
08:56:56 <Simon> https://bugzilla.gnome.org/show_bug.cgi?id=746937
08:57:20 <Simon> I don't know if it's safe to call xaccAccountSortSplits(acc, TRUE); at that time - I had to use the force option as the editlevel is non-zero
08:59:10 *** Jimraehl1 has left #gnucash
09:04:58 *** andy_ has quit IRC
09:09:20 *** Jimraehl1 has joined #gnucash
09:18:13 *** andy has joined #gnucash
10:24:59 *** StuM has quit IRC
11:07:15 *** GabrieleV_ has joined #gnucash
11:07:55 *** GabrieleV has quit IRC
11:07:55 *** GabrieleV_ is now known as GabrieleV
11:25:23 *** ErKa has joined #gnucash
11:27:45 *** GabrieleV_ has joined #gnucash
11:28:07 *** GabrieleV has quit IRC
11:28:07 *** GabrieleV_ is now known as GabrieleV
11:34:29 *** jeffsf has quit IRC
11:35:27 *** ArchangelXL has joined #gnucash
11:35:37 *** jeffsf has joined #gnucash
11:44:17 *** pacon has quit IRC
12:01:01 *** josePHPagoda has joined #gnucash
12:25:36 *** himaxx has joined #gnucash
12:30:34 *** ArchangelXL has quit IRC
13:11:15 *** jeffsf has quit IRC
13:22:05 *** jeffsf has joined #gnucash
13:37:06 *** fell__ is now known as fell
13:37:34 <fell> gjanssens: Nice to see you here again.
13:37:47 <fell> I have really missed you here.
13:39:35 <gjanssens> Glad to be back :)
13:39:55 <gjanssens> I was out of town for two weeks, and the rather busy catching up...
13:40:39 *** jeffsf has quit IRC
13:41:18 <fell> I have still the problem getting rid of the absolute path from make pot
13:43:10 <Simon> is it absolutely necessary to call checked_char_cast in sixtp-to-dom-parser.c? surely libxml2 returns an error on characters that are invalid in XML?
13:43:25 <gjanssens> It gets introduced if you do an out-of-tree build and then run make-gnucash-pot
13:43:37 <gjanssens> When building in-tree, it works
13:43:43 <gjanssens> (for me)
13:43:53 <fell> Hm
13:45:35 <fell> Any idea, what to tweak to get it working in a separate build dir?
13:46:33 <gjanssens> Unfortunately no. It has to do with the use of the $srcdir macro in the makefile, which is different for in-tree or out-of-tree builds
13:46:47 <gjanssens> We may have to add something like we do for the scm files
13:47:21 <gjanssens> Or cd into the src dir before calling xgettext may work also
13:47:46 <gjanssens> But I have to go now. I can test that part later
13:47:49 <gjanssens> Bye
13:47:58 <fell> OK, thx!
13:48:23 *** gjanssens has quit IRC
13:51:12 <Simon> about 25% of the cpu time for me when loading an XML is just calls to checked_char_cast or g_utf8_validate
13:51:33 <Simon> the progress bar is visibly faster if I remove the checks
13:55:14 <fell> bozonius: undo probably processed payments, open the invoice, unpost it, edit it, post it again, ...
13:59:20 *** fabior has quit IRC
14:04:43 *** fell_ has joined #gnucash
14:04:43 *** gncbot sets mode: +o fell_
14:13:34 <bozonius> fell: ay-yah-yah *headslap*
14:13:56 *** fell has quit IRC
14:13:59 <bozonius> but thanks; about what I was thinking also
14:16:28 *** jeffsf has joined #gnucash
14:22:32 *** fell_ is now known as fell
14:56:34 *** len has joined #gnucash
15:02:25 *** Krzysiek_K has left #gnucash
15:09:51 *** jralls has joined #gnucash
15:13:03 *** gncbot sets mode: +o jralls
15:13:44 <jralls> Back home, starting to build the release.
15:36:21 *** Leonarf has joined #gnucash
15:57:15 <warlord> wafflej0ck: FYI, using QIF you can absolutely specify both ends of transactions, and even specify split transactions.
15:57:49 *** GabrieleV_ has joined #gnucash
15:57:52 *** GabrieleV has quit IRC
15:57:52 *** GabrieleV_ is now known as GabrieleV
15:57:54 *** jeffsf1 has joined #gnucash
15:59:32 *** lmat has joined #gnucash
15:59:33 <warlord> jralls: welcome back
16:05:01 <jralls> warlord: Thanks. Now that the tarballs are on their way to SF, it's time to get groceries. BIAB.
16:05:09 *** jeffsf has quit IRC
16:05:09 *** jralls is now known as jralls_afk
16:19:18 *** isaacd has joined #gnucash
16:32:36 *** rubdos has quit IRC
16:56:36 *** ErKa has quit IRC
17:07:42 <wafflej0ck> warlord, thanks good to know too, I'll probably end up doing some things with QIF as well, but for the time being will probably just try to figure out how to integrate the invoices and payments, as is I just setup a separate DB where I post invoices for customers entered in our CRM and they can sign into the site to make payments, but none of that goes into GNUCash automatically yet I still have to enter the invoice and payment
17:07:43 <wafflej0ck> in GNUCash as well to keep our records straight
17:12:58 *** MechtiIde has quit IRC
17:15:38 *** misha has joined #gnucash
17:20:14 *** misha__ has joined #gnucash
17:24:22 *** misha_ has quit IRC
17:28:46 *** misha has quit IRC
17:29:26 *** ErKa has joined #gnucash
17:30:48 *** Gbarr has joined #gnucash
17:43:24 *** misha__ has quit IRC
18:20:29 *** Leonarf has left #gnucash
18:26:49 *** misha has joined #gnucash
18:41:58 *** jralls_afk is now known as jralls
18:50:40 <Simon> the checked_char_cast() in sixtp-to-dom-parser.c looks completely unnecessary
18:51:00 <Simon> when I test it loading files with characters that aren't accepted, it returns an error instead of being able to replace them with "?"
18:51:12 <Simon> so it's just increasing the load time with no benefit
18:57:57 <jralls> Simon: That's interesting. What OS and GLib version?
18:58:29 <Simon> Ubuntu 14.10 and 2.42.1-1~ubuntu1
18:58:51 <Simon> I'm getting a Guile stack trace instead of "?"s in the strings
18:59:23 <jralls> A *guile* stack trace? That's weird. What is it?
18:59:29 <Simon> http://s85.org/iDfOY0xL
19:01:38 <Simon> actually I might have been testing with a version that has the check removed...
19:02:42 <jralls> Ah. So your bad UTF8 gets through to guile. That would make more sense.
19:03:05 <Simon> is the checking strictly necessary given that gnucash does it on write?
19:03:45 <Simon> my load time seems to be based on two factors: reading the XML, and opening all the account tabs...
19:04:21 <Simon> it might actually be the latter that takes the most time but oprofile is showing up checked_char_cast
19:04:26 <jralls> Simon: Yes, it's possible for the file to be read incorrectly, especially if it's been compressed.
19:08:36 <jralls> Yes, it's known that checked_char_cast slows down file loading by a huge amount. It's also known that corrupted strings are rare and that the only known case in the wild came from an OFX import. The problem is that it doesn't cause that crash until the next time you load the file with the bad UTF8 in it, so we're being exceedingly cautious to avoid locking users out of their files.
19:08:50 <Simon> also, does it make sense to load splits in reverse of the file order when the file is sorted?
19:09:42 <Simon> right, but you'd still have checked_char_cast on write so you couldn't write a bad file
19:09:49 <Simon> for some reason writing isn't slow
19:10:30 <Simon> src/engine/Account.c: priv->splits = g_list_prepend(priv->splits, s);
19:11:15 <Simon> glib doesn't have doubly-linked lists? :/
19:11:30 <Simon> or at least a "last" pointer
19:12:40 <Simon> "Reverses a GList. It simply switches the next and prev pointers of each element." this is silly :?
19:12:42 <Simon> :/
19:14:01 <jralls> Simon: GList is a doubly-linked list. It's used widely in GnuCash. No, it has only a begin pointer. Yes, much of GLib is silly. That's part of why we're rewriting everything in C++, where the preferred container is std::vector. *Much* faster than lists.
19:14:45 <Simon> the whole point of something like glib is that they could just add an end pointer and everything would get a benefit :|
19:15:25 <jralls> Simon: I don't think that the GLib devs think that's the point of the library at all.
19:16:07 <Simon> I can see that from reading the mailing list
19:16:31 <Simon> where they seem to think reversing it is an efficient idea
19:18:06 <Simon> oh there is no head either
19:18:27 <Simon> couldn't it just set the newly added item as the head?
19:19:23 <Simon> oh, right
19:19:30 <Simon> the whole thing is still in reverse order
19:19:33 *** pacon has joined #gnucash
19:22:27 <jralls> Simon: If you want to debate GLib's (de)merits with someone, #gtk+ might be more productive.
19:36:25 *** jralls is now known as jralls_afk
20:13:22 *** himaxx has joined #gnucash
20:18:34 *** himaxx has joined #gnucash
20:19:16 *** himaxx has quit IRC
20:31:01 *** misha has quit IRC
20:33:35 *** misha has joined #gnucash
20:38:56 *** misha has quit IRC
20:55:11 *** misha has joined #gnucash
21:32:20 *** warlord2 has joined #gnucash
21:35:35 *** warlord has quit IRC
22:09:34 *** warlord2 has quit IRC
22:27:42 *** warlord has joined #gnucash
22:27:42 *** gncbot sets mode: +o warlord
22:50:51 *** ArchangelXL has joined #gnucash
23:05:25 <wafflej0ck> should the 2.6.6 tag be compiled with guile 2.0?
23:07:36 <wafflej0ck> I have 2.6.1 that came from the ubuntu trusty repos, and 2.6.5 that I compiled myself but when doing that I removed whatever version of guile I had and added 2.0 because that's what it asked for during the autogen.sh or the configure script (I forget which one), but the receivable aging report is throwing an error it seems regardless of if I use the 2.6.5 or 2.6.1 versions so just grabbed the 2.6.6 tag and compiling it now but
23:07:36 <wafflej0ck> not sure if I'm still going to have the same problem (PS tried going back to my regular books.gnucash file based stuff since I've been messing around with the DB but still see the error for that report)
23:08:34 <wafflej0ck> it seems to be an out of range error I can post the message I see if I have --debug flag set when running it
23:08:51 <wafflej0ck> if someone can help and needs any info let me know
23:14:42 <wafflej0ck> here's the --debug output http://pastebin.com/K3avan7A seems to still be a problem, it doesn't seem to come up in any of the other reports (makes some sense since I can see it's something in the aging.scm file but I can't fully decipher what's gone wrong)
23:51:57 *** jlo has joined #gnucash
23:57:08 *** lwells has quit IRC
23:57:35 *** lwells has joined #gnucash