2008-08-28 GnuCash IRC logs
00:05:10 *** twunder has quit IRC
02:01:44 *** Rolf has joined #gnucash
02:06:57 *** Zoolooc_ has joined #gnucash
02:22:03 *** esperegu has joined #gnucash
02:31:42 *** Coweater has quit IRC
02:45:37 *** _gunni_ has joined #gnucash
02:50:27 *** ErKa has joined #gnucash
02:51:36 *** localhost1 has joined #gnucash
02:53:46 *** |gunni| has quit IRC
02:58:54 *** localhost has quit IRC
03:08:46 *** Zoolooc_ has quit IRC
04:09:23 *** ErKa has quit IRC
04:11:44 *** aindilis` has quit IRC
04:14:56 *** aindilis` has joined #gnucash
04:22:15 *** Coweater has joined #gnucash
05:01:36 *** ErKa has joined #gnucash
05:33:37 *** fabi has joined #gnucash
06:02:38 *** Ozaed__ has joined #gnucash
06:04:18 *** Mansig has joined #gnucash
06:11:31 *** Ozaed_ has quit IRC
06:12:11 *** ErKa has quit IRC
06:34:55 *** ErKa has joined #gnucash
06:59:42 *** andi5 has joined #gnucash
06:59:43 *** gncbot sets mode: +o andi5
07:08:54 *** ErKa has quit IRC
07:19:58 *** esperegu has quit IRC
07:20:47 *** ErKa has joined #gnucash
07:45:34 *** esperegu has joined #gnucash
08:08:02 *** ErKa has quit IRC
08:11:46 *** martincleaver has joined #gnucash
08:32:20 *** JimRaehl has left #gnucash
08:35:29 *** fabi has quit IRC
09:12:43 *** JimRaehl has joined #gnucash
09:34:43 *** Mike has joined #gnucash
09:40:04 *** Mike has left #gnucash
09:41:46 *** twunder has joined #gnucash
09:42:04 *** ErKa has joined #gnucash
10:11:45 *** donfede has joined #gnucash
10:49:31 *** Mike has joined #gnucash
10:50:25 *** Mike has left #gnucash
11:15:44 *** esperegu has quit IRC
11:34:13 *** warlord-afk is now known as warlord
11:46:12 *** ErKa has quit IRC
12:04:34 *** ErKa has joined #gnucash
12:04:56 *** nomeata has joined #gnucash
12:19:22 *** micha has joined #gnucash
12:19:57 <micha> Hi andi5
12:20:04 <andi5> hi micha
12:20:39 <micha> do you have any idea on how to implement the share certificate cache for AqBanking?
12:21:19 <andi5> martin wrote about it on the mailing list as response to my question... i cannot search for it right now though
12:21:59 <micha> do you remember which mailinglist?
12:22:50 <micha> got it
12:22:53 <andi5> oh, fine :)
12:24:20 <micha> he wrote that his code examples only work if our GwenGui supports certificate caches
12:24:39 <micha> I tend to answer "yes", but I am not sure
12:26:14 <andi5> what does "support certificate caches" mean?
12:26:45 <micha> yes, that's my very question :-}
12:26:57 <andi5> which mail do you see? do you have a link for me?
12:27:58 <micha> http://www.mail-archive.com/aqbanking-devel@lists.sourceforge.net/msg02703.html
12:29:13 <andi5> ich denke nicht, aber er hat ja eine schöne code-referenz angegeben :)
12:29:39 <andi5> ich weiß nicht, wann ich dazu käme, ich schaffe es ja nicht einmal, mich um backports zu kümmern...
12:29:46 <andi5> oops, english
12:30:40 <micha> maybe we should ask Martin to clarify what "support certificate cache" means
12:33:11 <andi5> well, take a look at the setDbCerts function of his cgui implementation
12:33:18 <warlord> I suspect it means we need to remember the SSL certificates.
12:33:48 <andi5> i.e. we read after initing online, and save back when finishing
12:34:11 <andi5> meanwhile we can implement the cache the way we like it, as long as we manage to convert the data appropriately... that is only guessed though
12:35:15 <andi5> whenever we know how to extract the relevant data, we may prefer working with the provided structures directly
12:40:39 *** nomeata has quit IRC
12:47:52 *** andi5 has quit IRC
12:49:19 <micha> Is there any way to minimize the relinking all the time if I only change very little code?
12:51:34 <warlord> If you only change .c files then you only need to rebuild (and reinstall) in those sub-dirs.
12:51:44 <warlord> If you change a .h then you need to rebuild everything.
12:59:10 <micha> right now I change only .c files, but on "make install" it relinks a huge bunch of libraries again
12:59:38 *** tats has joined #gnucash
12:59:45 <tats> hello everyone
13:02:43 <warlord> micha: as I said, only run make and make install in the directories you change.
13:02:55 <micha> ah, got you wrong
13:02:58 <micha> thanks
13:04:21 <warlord> you're welcome
13:21:52 <micha> I'm currently investigating http://bugzilla.gnome.org/show_bug.cgi?id=549595
13:21:59 <micha> and I discovered the problem
13:22:40 <micha> in http://svn.gnucash.org/trac/browser/gnucash/trunk/src/backend/file/gnc-backend-file.c#L400 link() returns ENOSYS
13:24:15 <micha> I could easily add ENOSYS in an #ifdef clause like EOPNOTSUPP is, but I'm somewhat curious for the reasons why only EOPNOTSUPP get cought by the code
13:25:26 <micha> shouldn't there be an else clause in line 417 letting Gnucash explode if something unknown (like this sshfs bug was up to now unknown) happens?
13:25:33 <warlord> Probably because ENOSYS isn't listed as an expected error from link(2)
13:26:15 <micha> yes, but what's about all the other *documented* reasons link() might fail for?
13:26:51 <micha> the code right now doesn't care...
13:27:10 <warlord> If it fails for those other reasons then it probably can't copy the file either.
13:27:19 <micha> in this case it can
13:27:54 <micha> and why not giving copy_file() a chance?
13:28:42 <warlord> Because if the error is EFAULT, EIO, ELOOP, EMLINK, ENAMETOOLONG, ENOENT, ENOMEM, ENOSPC, ENOTDIR, EROFS -- the copy_file will fail too.
13:29:20 <warlord> We should probably catch EXDEV... But that error generally can't happen in our usage because you can't split a single directory across multiple devices.
13:31:25 <warlord> Now, I think it's a bug that we're conflating err_ret and the return from copy_file.. copy_file returns a gboolean, whereas err_ret is an integer..
13:32:07 <warlord> But that's a different issue.
13:32:20 <micha> yes, and a rather cosmetical one..
13:32:38 <warlord> Well, not purely cosmetic -- it's causing us NOT to report failures.
13:33:24 <micha> wouldn't it be safe, to fail over to copy_file() *always* in case link() failed?
13:34:12 <warlord> Well you could lose the errno setting...
13:34:33 <warlord> And as I said, in the MAJORITY of errors the copy_file will also fail, so why even try it when you know it will fail?
13:35:22 <warlord> IMHO: We should fix the return codes so that we report the error even when we don't call copy_file.
13:35:25 <micha> the benefit would be having all link() failures caught at once.
13:35:43 <warlord> .. and then you can add ENOSYS to the list of "try copy_file anyways" errors.
13:36:26 <warlord> So what I would do is add a "gboolean copy_err = FALSE;" and then have copy_file() set copy_err within the same logic now..
13:36:46 <warlord> Then change if(!err_ret) to if(!copy_err)
13:36:56 <micha> "gboolean link_err" you mean?
13:37:00 <warlord> No......
13:37:17 <warlord> err_ret is the return from link()
13:37:24 <micha> ah, ok
13:37:46 <warlord> no need to change that.
13:38:10 <micha> sure, got it
13:38:24 <warlord> Or... you can add: else { err_ret = 0; }
13:38:28 <warlord> that would work too.
13:38:40 <warlord> But I think adding the new variable is cleaner.
13:45:37 *** ErKa has quit IRC
13:48:21 <micha> but if link() returns one of the other error codes the function should return FALSE too, right?
13:49:42 *** esperegu has joined #gnucash
13:56:34 <warlord> And it will given my proposed changes.
13:57:36 <micha> without adding an explicit else clause?
13:57:59 <micha> ah, sure
13:58:03 <warlord> Yes, because you change the test from !err_ret to !copy_err and you initialize copy_err
14:01:29 <micha> maybe it should be copy_success and initialized to TRUE to avoid headaches
14:02:00 <micha> this would better reflect the return result of copy_file()
14:02:41 <micha> oops, no, still initialized to FALSE
14:06:17 <micha> I'm a little uncomfortable with all these implicit assumptions, but you're right: it will work.
14:20:16 <micha> This is the result: http://bugzilla.gnome.org/attachment.cgi?id=117542&action=view
14:21:08 <warlord> That looks good.
14:21:11 <warlord> Does that fix the problem?
14:21:22 <micha> I've checked it: With this patch applied Gnucash works now over sshfs without data loss
14:22:07 <warlord> Cool.
14:25:43 <micha> should be backported too ;)
14:29:06 <micha> bye
14:29:11 *** micha has quit IRC
15:14:23 *** sjc has joined #gnucash
15:17:54 *** |gunni| has joined #gnucash
15:17:58 *** esperegu has quit IRC
15:23:58 *** piksi has joined #gnucash
15:24:30 <piksi> hi people, is there a version of gnucash that would run solely on a web interface? or possibilities to run it in such way?
15:24:41 <jsled> no.
15:25:17 <piksi> darn. thanks anyway. i'm looking for a sql/www based accounting software
15:26:42 *** _gunni_ has quit IRC
15:27:30 <warlord> piksi: try sql-ledger (or whatever it's called now)
15:28:32 <piksi> thanks for the hint, appreciated
15:35:53 *** piksi has left #gnucash
17:13:03 *** twunder has quit IRC
17:44:15 *** tats has quit IRC
17:58:11 *** BlackBsd has joined #GnuCash
18:04:35 *** todd_ has joined #gnucash
18:05:47 *** BlackBsd has joined #GnuCash
18:08:37 *** todd has quit IRC
18:09:52 *** BlackBsd has quit IRC
18:22:33 *** donfede has quit IRC
18:45:42 *** cortana has joined #gnucash
19:05:35 *** rog1 has joined #gnucash
19:10:13 *** cortana has quit IRC
19:15:28 *** Rolf has quit IRC
19:27:36 *** jcreigh has joined #gnucash
19:39:30 *** warlord is now known as warlord-afk
19:52:24 *** ErKa has joined #gnucash
20:14:12 *** JimRaehl has left #gnucash
20:15:20 *** JimRaehl has joined #gnucash
20:58:01 *** ErKa has quit IRC
21:00:44 *** sjc has quit IRC
22:39:56 *** benoitg has joined #gnucash
22:43:10 *** todd_ has quit IRC
22:43:22 *** todd has joined #gnucash
23:18:20 *** jcreigh has quit IRC
23:38:41 *** ben_goodger has quit IRC
23:39:26 *** ben_goodger has joined #gnucash