2007-02-13 GnuCash IRC logs

00:14:03 *** motin has joined #gnucash
00:16:48 *** dbr has quit IRC
01:27:47 <prock> it compiled! :)
01:58:15 *** Esaj has joined #gnucash
02:01:30 *** jpeach has joined #gnucash
02:08:48 *** warlord is now known as warlord-afk
03:47:12 *** RallyU has joined #gnucash
03:50:59 *** ErKa has joined #gnucash
04:53:32 *** RallyU has quit IRC
05:40:41 *** jpeach has quit IRC
07:21:14 *** cstim has joined #gnucash
07:21:14 *** gncbot sets mode: +o cstim
07:23:06 <cstim> @tell andi5 Indeed this generated header issue caused aqbanking's "make check" to crash. Check the difference between $gwen_prefix/include/gwenhywfar/system.h and $tmpdir/gwenhywfar-2.5.3/gwenhywfar/system.h - the header that was used during compiling gwen erroneously thought this wasn't windows.
07:23:06 <gncbot> cstim: The operation succeeded.
07:24:01 *** ceplma has joined #gnucash
07:30:32 <cstim> @tell andi5 Once I made sure gwen was being compiled with the correct headers, "make check" in aqbanking ran again.
07:30:32 <gncbot> cstim: The operation succeeded.
07:34:35 *** twunder has joined #gnucash
08:04:02 <jsled> Today's "gnucash-devel Moderation Queue: Spam or Ham?!" ... From: anas@chartered-dubai.com, Subject: 1 question
08:04:40 * cstim votes for spam
08:05:33 <jsled> Ooh, and a super-special followup bonus question in today's episode: From: zvd@itechforbiz.com, Subject: This file is usually updated about every two minutes.
08:06:31 <cstim> hm... not ham. Maybe minced meat?
08:06:34 <jsled> heh.
08:07:02 <jsled> On the first one, I too thought spam ... we were both wrong.
08:07:17 <cstim> interesting.
08:07:19 <jsled> The second is spam, however.
08:08:55 <cstim> the message itself is ... strange. Someone who has *a* Gnucash.
08:40:40 <cstim> Is prock == Peter McAlpine?
08:40:53 <jsled> yes.
08:41:25 <cstim> I'm building with his Makefile patch on windows, and it fails. I'll email some feedback soon.
08:48:12 *** motin has quit IRC
08:48:50 *** spel has joined #gnucash
08:50:13 *** warlord-afk is now known as warlord
08:50:42 <warlord> It's possible that some things that gnucash builds as modules really should be built as shared libraries..
08:56:05 *** ServerCrash has joined #gnucash
08:57:02 <ServerCrash> hi, where can i get more fancy invoice templates ?
08:57:36 <ServerCrash> like i would like to add logo, and restructure some of the content organisation in fancy invoice template
08:57:41 <ServerCrash> anyone with any idea
08:59:47 <warlord> the invoices aren't templates.
09:00:20 <warlord> you can play with the gnucash stylesheets.
09:01:45 <ServerCrash> warlord, ya working on there only
09:02:11 <ServerCrash> warlord, can you name any other small business accounting application that is as good as gnucash or better ?
09:02:20 <warlord> Nope.
09:02:27 <warlord> I dont know of anything better (and free)
09:06:37 *** ServerCrash has quit IRC
09:09:01 <jsled> Ah ha! When I send messages, I send "...jsled" (with a valid signature); when they get delivered back to me, it's "....jsled"! Something's doing a bad thing.
09:11:08 <warlord> It's on your end..
09:11:10 <warlord> I see:
09:11:11 <warlord> ...jsled
09:11:18 <jsled> Huh?
09:11:20 <warlord> (in my inbox)
09:11:37 <warlord> My guess: the IMAP client or MUA.
09:12:01 <jsled> Sure enough, that's what's in the list archives. :/
09:12:26 * cstim also sees three dots.
09:12:58 <warlord> So anything that's doing it is after it reaches asynchronous
09:13:07 <cstim> prock: these linker flags you've removed really don't mean *static* linking. Why did you think they did?
09:42:39 <warlord> cstim: I wonder if we should continue to think about what parts are libraries and what are modules..
09:45:39 <cstim> sure.
09:47:01 <jsled> How much work is it to fix it? It seems like a fair bit of re-work, though it'll mostly be straightforward once the first one is understood.
09:48:40 <warlord> I dont know.
09:49:57 <warlord> It depends if we're actually using the GNCmod initialization routines for those libraries..
10:00:09 <warlord> As I look, that patch doesn't make sense..
10:00:30 <warlord> there should be many more issues than just those.
10:10:25 *** warlord is now known as warlord-afk
10:20:04 *** jpeach has joined #gnucash
10:20:07 *** jpeach has left #gnucash
10:39:24 <prock> cstim: while they may not explicitly mean "statically link these libraries" it's what ended up happening: http://paste.debian.net/21859 . libgncmod-register-core.dylib gets added to the link line for libgncmod-register-gnome.dylib
10:41:25 <prock> btw my patch was definately not meant as a "this is the for-real solution" patch... I just wanted to show a way to get things running on the MacIntel
10:41:28 *** writerz_ is now known as exper
10:42:40 <cstim> prock: sure.
10:43:50 <cstim> prock: are you saying that if the libfoo.dylib is added to the link line, libfoo will be statically linked into the result?
10:44:27 <prock> cstim: that was what I was saying because I don't see any other reason that this error would be happening unless that were the case.
10:45:16 <cstim> But all the (non-gnucash) gnome/glib libraries are also mentioned on the link line: liblib-2.0.dylib etc.. Those are surely not linked in statically, are they?
10:46:01 <prock> I suppose I'm seeing gnc modules and shared libraries as the same thing when from what you have suggested they are not.
10:46:26 <cstim> I don't have any doubts that your finding is very relevant to the MacIntel build problems, but I'm just confused about the term "statically linked" because that is a very specific thing.
10:47:49 <cstim> No, you are right - from the point of view of the linker, gnc modules and shared libraries are the same thing. (They are not, when talking about some of their initialization, but that's irrelevant when talking about the linker line, IIRC.)
10:48:07 <prock> If you feel I'm incorrectly throwing around the term "statically linked" then it wont be hard to convince me ; I have very little experience with libtool and building on the mac
10:48:37 <prock> but I don't know how else to interpret the linker line (pasted above) except that it seems to be statically linking the dylibs together.
10:49:10 <prock> otherwise why would register-core need to be on the linker line... it should just be able to be pulled in via "-l"
10:49:55 <cstim> Do you have "nm" available? On Linux you could run "nm -C .libs/libgncmod-register-gnome.so" and see all symbols that are defined inside that library.
10:50:45 <prock> what would I nm? It can't build libgncmod-register-gnome (.dylib on the mac)
10:51:39 <cstim> just any other dylib.
10:51:50 <cstim> libgncmod-register-core.dylib?
10:52:08 <prock> one sec
10:52:16 <cstim> example output of nm -C libgncmod-register-gnome.so |grep gnc_
10:52:17 <cstim> is here http://pastebin.ca/353760
10:52:53 <cstim> you see some lines have a "U" and no address, whereas others have a "T" or "t" and an address?
10:53:10 <cstim> (I'm trying to explain the term "statically linking" to you)
10:55:00 *** dbr has joined #gnucash
10:55:08 <prock> I see, "U" (undefined) is for dynamically linked symbols?
10:55:37 <cstim> exactly. Whereas all "T" and "t" are statically linked symbols - basically, all functions that are *really* inside this object file (=library).
10:55:39 <prock> lower case "t" is variables whereas upper case "T" are functions?
10:56:22 <cstim> no, lower case means it isn't exported whereas upper case means it is exported (=can be called from outside).
10:57:00 <cstim> but anyway. Everything that appears here with "U" means the linker has dynamically linked to the library where the function really exists.
10:57:31 <prock> ah... so two linked libs may not share any symbols that have "T" but they may both contain "t" symbols.
10:57:36 <prock> is this correct?
10:57:44 <cstim> yes, that as well.
10:58:46 <cstim> But back to "statically linked": I don't think that a "libfoo.dylib" on the linker line means that libfoo has been linked into statically, because that would mean all of libfoo's symbols will appear as "T"/"t" in the resulting object file. This is probably not the case.
10:59:07 <cstim> Rather, a "libfoo.dylib" on the linker line means libfoo has been linked to *dynamically*.
10:59:21 <cstim> prock: can you follow?
10:59:50 <dbr> cstim: my attempt to build gwen 1160 gets farther than before, but it's complaining about not being able to find regex. I've added several of the /c/soft/regex dirs to PATH, and rerun config, but it's still failing make
11:01:47 <cstim> dbr: my gwenhywfar-SVN build has the following configure line:
11:02:02 <cstim> ./configure LDFLAGS=-L/c/work/gnucash/regex/lib --with-openssl-includes=/c/work/gnucash/openssl/include --with-openssl-libs=/c/work/gnucash/openssl/lib --prefix=/c/work/gnucash/gwen
11:02:39 <cstim> and the "make" command doesn't set another LDFLAGS; it's just plain "make".
11:03:12 <cstim> dbr: do you have such a LDFLAGS in gwen's configure?
11:03:17 <cstim> install.sh has it.
11:05:04 <dbr> cstim: I didn't have the LDFLAGS set for regex. I'll rerun configure. Thanks.
11:06:55 <cstim> dbr: what windows version do you have?
11:07:08 <dbr> winXP SP2
11:07:39 <prock> cstim: yes I follow thanks. I had to install a better 'nm' because it seems that apple's is very old.
11:09:13 <prock> the offending symbols are "B" (in the uninitialized data section (known as BSS)"
11:09:54 *** warlord-afk is now known as warlord
11:09:56 <prock> http://paste.debian.net/21860
11:10:18 *** ceplma has quit IRC
11:10:50 <cstim> gnc_module_age is offending the MacIntel Linker?
11:11:02 *** dbr is now known as dbr-afk
11:11:22 <cstim> oops. Every "libgncmod-foo" (with the "mod", i.e. a module) has to contain this symbol.
11:12:05 <cstim> As warlord said, we need to refactor the code so that the stuff from libgncmod-register-code isn't in a module anymore but instead a library.
11:12:53 <prock> the offending symbols: _gnc_module_current _gnc_module_init _gnc_module_path _gnc_module_revision _gnc_module_system_interface _gnc_module_age _gnc_module_description
11:13:28 <prock> but that is just when building register-gnome. There were a number of other failures (usually the same symbols) but I didn't keep logs of each failure.
11:13:30 <cstim> ok. I'll hand this off to {warlord,jsled,andi5}
11:13:43 <prock> ok
11:13:58 *** ErKa has quit IRC
11:14:13 <warlord> prock: what I dont understand is why you can't link to libgncmod-register-* but you CAN link to libgncmod-engine
11:14:37 <prock> yes I noticed that last night as well...
11:15:35 <warlord> i'll point out to andi5 that this linker issue is why the symbols originally had the LTX_<module>_<symbol>
11:16:03 <prock> yeah, that's why the 2.0 branch builds ok.
11:16:31 * cstim uploaded r15574 win32 build.
11:20:18 <warlord> prock: what I dont understand is why it builds on every other platform, including Mac/PPC!
11:29:53 <dbr-afk> cstim: get the flags right, and it works. How 'bout that? Now, do I 'make check' and not bother with make install if it fails?
11:30:35 *** dbr-afk is now known as dbr
11:33:51 <cstim> dbr: yes
11:34:08 *** ErKa has joined #gnucash
11:34:15 <dbr> update: if I run 'make check' from the top svn dir, it fails because libssl32.dll is not found
11:34:46 <cstim> sure. Add /c/soft/openssl/bin to PATH (install.sh does it).
11:36:49 <dbr> hmm. Did I miss a couple lines setting up /etc/profile.d/install.sh? Many of the directories did get added, but not openssl. sigh.
11:38:34 <dbr> OK. make check passes all tests for gwen 1160. I'll do aqbanking again next. but first it's off to the office until I almost get snowed in.
11:40:32 *** dbr has quit IRC
11:43:55 *** cstim has quit IRC
11:52:18 *** benoitg has joined #gnucash
11:52:49 *** benoitg has left #gnucash
12:11:10 *** andi5 has joined #gnucash
12:11:11 *** gncbot sets mode: +o andi5
12:12:12 <cortana> could someone check
12:12:21 <cortana> #347623 -- the fix should be some simple changes to the documentatino files
12:14:46 <warlord> cortana: I think this has already been fixed in the docs sources.
12:14:54 <cortana> hm
12:15:08 <cortana> i only have gnucash-docs 2.0.1 installed
12:15:51 <cortana> does gnucash have a websvn viewer or similar
12:15:58 <cortana> nm i found it :)
12:23:33 <cortana> there are still other missing figures
12:28:26 <warlord> Oh? Even in the daily builds?
12:31:05 <andi5> warlord: WRT LTX_module_symbol .... you know who was responsible for removing these prefixes :) .... that is why i wanted to test my branch on at least win32 and macos... but well, i did not expect changes between macos/ppc and macos/intel...
12:31:39 <warlord> andi5: yes, I know who made those changes.. and I know why you wanted to test it...
12:31:59 <warlord> I didn't expect differences between macos/ppc and macos/x86 either
12:32:25 <andi5> oh, actually i mean who was responsible for removing the prefixes while loading the modules :) ... i.e. libltdl =)
12:32:56 <warlord> Oh..
12:33:20 <warlord> Well, this is why I was trying to state hard that you should never link against a GNCmodule.
12:33:47 <andi5> hey, it was not me ;-9
12:33:59 <warlord> hehe
12:34:56 <andi5> prock: so what linker do you use? i am sorry, i did not read that.... is it the gnu ld?
13:04:34 <prock> ld is provided by mac
13:04:55 <prock> (XCode) "ld - Mach object file link editor"
13:27:26 <andi5> prock: is there some "gnu linker" you can test with?
13:28:58 <warlord> andi5: i dont think so...
13:29:26 <andi5> warlord: do you use the same on macos/ppc?
13:30:59 <warlord> I'm pretty sure.
13:32:53 <warlord> I only see /usr/bin/ld which is from XCode.
13:33:09 <andi5> ok... so what is the difference....
13:33:41 <warlord> I dont know.
13:35:29 *** exper_ has joined #gnucash
13:40:52 <andi5> @tell cstim yeehaw.... see http://pastebin.ca/raw/354012 .... seems like we have found a definite problem :) ... congratulations! .... let us see what happens if i rerun configure and make install in gwen...
13:40:52 <gncbot> andi5: The operation succeeded.
13:41:28 *** franz has joined #gnucash
13:43:02 *** exper has quit IRC
13:57:30 *** MrN has joined #gnucash
13:58:25 <MrN> hi
14:01:34 *** mnoir has joined #gnucash
14:05:44 *** |gunni| has joined #gnucash
14:11:24 *** andi5 has quit IRC
14:31:55 *** mnoir has quit IRC
14:32:50 *** sjc has joined #gnucash
14:37:11 *** jpeach has joined #gnucash
14:37:22 *** jpeach has left #gnucash
14:43:28 *** sjc_ has joined #gnucash
14:43:37 *** spel has quit IRC
14:44:41 *** sjc_ has quit IRC
14:45:03 *** sjc has quit IRC
14:45:41 *** sjc has joined #gnucash
14:55:16 *** andi5 has joined #gnucash
14:55:16 *** gncbot sets mode: +o andi5
15:00:03 *** mnoir has joined #gnucash
15:07:59 <andi5> @tell cstim so after the second configure run and recompiles of gwen&aq, the worlds looks a lot better.... *happily awaiting ++gwen*
15:07:59 <gncbot> andi5: The operation succeeded.
15:09:47 <warlord> andi5: when we do the g_module_symbol() I wonder if we could dynamically lookup the symbol name based on the name of the .so? I have no idea if that would help/work or solve the mac/x86 link issue.
15:11:31 <andi5> that c/would help... well, it is all in gnc-module.c..... but do we really need this?
15:14:28 * andi5 is kinda angry that macos/intel does not work... it is too similar to a supported platform from my point of view...
15:15:03 <andi5> dbr is ppc-only, right?
15:17:10 <warlord> Yes, I think that dbr is ppc-only.
15:18:00 *** franz has quit IRC
15:34:25 *** cstim has joined #gnucash
15:34:26 *** gncbot sets mode: +o cstim
15:34:34 <cstim> hi
15:34:34 <gncbot> cstim: Sent 1 hour and 53 minutes ago: <andi5> yeehaw.... see http://pastebin.ca/raw/354012 .... seems like we have found a definite problem :) ... congratulations! .... let us see what happens if i rerun configure and make install in gwen...
15:34:35 <gncbot> cstim: Sent 26 minutes ago: <andi5> so after the second configure run and recompiles of gwen&aq, the worlds looks a lot better.... *happily awaiting ++gwen*
15:35:09 <cstim> @tell andi5 Good to hear. What about make check? For me it still fails on winXP+aqbanking.
15:35:09 <gncbot> cstim: The operation succeeded.
15:35:18 <andi5> .
15:35:18 <gncbot> andi5: Sent just now: <cstim> Good to hear. What about make check? For me it still fails on winXP+aqbanking.
15:36:09 <andi5> hm... install.sh checked automagically, right? ... then they passed
15:37:09 <andi5> ./testlib foo .... works
15:37:25 <andi5> cancelling aq setup druid... works
15:37:39 * andi5 is happy
15:37:42 <cstim> no, install.sh did not check in aqbanking.
15:37:57 <andi5> oh, you are right
15:38:06 <andi5> (worked nonetheless)
15:39:30 <cstim> it did? it didn't for me :-(
15:40:10 <cstim> (on winXP. on win2000 all is fine.)
15:40:27 <andi5> oh.... wrong directory (why do all testlibs have the same name ;-))
15:40:45 <cstim> yeah. ./testlib foo in src/frontends/cbanking?
15:40:57 <andi5> it does not print anything ... anymore
15:41:56 <cstim> it looks as if it hangs? that's what I saw in MSYS as well. In cmd.exe it should work.
15:44:29 <andi5> "exetype .libs/testlib.exe console" might help
15:45:33 <andi5> or put differently: it helped me :)
15:46:11 <andi5> why the heck does this program beep? is that intended?
15:48:10 <cstim> I think it is, so that you recognize the PIN question. Complaints > aqbanking-devel.
15:48:30 <andi5> hehe
15:49:28 <cstim> I'll be out of town (=afk) the next two days.
15:49:53 <andi5> what about your addiction? i hope it will not cool down ;-)
15:50:02 <cstim> and actually I'm wasting way too much of my thesis writing time on gnucash/win32. I *must* stop.
15:50:19 <cstim> I can stop *anytime*. I tell you.
15:50:23 <andi5> oops... actually i thought you had just finished *g*
15:50:33 <andi5> yeah, i know
15:50:36 <andi5> just this one more time
15:52:19 * andi5 will give f::q a shot the following days.... getting our guile code to play nicely with non-ascii file paths seems like a major headache i do not need right now
15:54:03 <warlord> andi5: wrap g_spawn and use that instead. ;)
15:54:23 <andi5> warlord: _what_ to spawn.... we need perl and f::q perl modules...
15:55:33 <warlord> Spawn gnc-fq-helper
15:56:45 <andi5> i am not yet sure this is the way to go.... we will see
15:57:08 *** mnoir has quit IRC
15:58:09 <andi5> oh, -helper probably, but maybe not -update...
15:58:48 *** Jack has joined #gnucash
15:58:52 * andi5 will report back later...
15:59:10 *** andi5 has quit IRC
15:59:57 <warlord> @tell andi5 GnuCash only needs to call -check and -helper...
15:59:57 <gncbot> warlord: The operation succeeded.
16:00:21 *** Jack has left #gnucash
16:06:30 *** Esaj has quit IRC
16:10:44 *** cstim has quit IRC
16:14:52 *** Esaj has joined #gnucash
16:42:30 *** benoitg has joined #gnucash
16:43:26 *** benoitg has left #gnucash
17:03:24 *** twunder has quit IRC
18:07:33 *** jpeac1 has joined #gnucash
18:07:37 *** jpeac1 has left #gnucash
18:09:38 *** ErKa has quit IRC
18:35:26 *** warlord is now known as warlord-afk
18:38:58 *** twunder has joined #gnucash
19:00:33 *** MrN has quit IRC
19:41:50 *** esodan has joined #gnucash
20:07:15 *** |gunni| has quit IRC
20:19:54 *** esodan has quit IRC
20:19:58 *** cortana has quit IRC
20:20:28 *** cortana has joined #gnucash
20:26:01 *** Esaj has quit IRC
21:10:52 *** hampton|away is now known as hampton
21:11:34 <hampton> I think the g_spawn approach should work. gnc-fq-check and gnc-fq-helper are the two perl scripts called by gnucash.
21:16:54 *** prock_ has joined #gnucash
21:25:32 *** prock has quit IRC
21:25:38 *** sjc has quit IRC
21:25:46 *** prock_ is now known as prock
21:43:55 *** Wilddev has joined #gnucash
21:43:55 *** gncbot sets mode: +o Wilddev
21:51:00 *** spel has joined #gnucash
21:52:17 <jsled> hey Wilddev
21:55:10 <Wilddev> hi jsled
21:59:45 *** spel has quit IRC
22:02:06 *** warlord-afk is now known as warlord
22:02:07 <warlord> hey Wilddev
22:04:53 <Wilddev> hi warlord
22:05:06 <Wilddev> looool
22:06:06 <warlord> LOL!
22:06:21 <warlord> jsled: agreed.
22:07:02 * jsled goes afk.
22:07:13 <Wilddev> jsled: so what's sx=cleanup?
22:07:50 <Wilddev> warlord: [off] Yeah I got that part its just bill was being so misterious in his email
22:08:11 <Wilddev> hmm didnt do that right lol
22:08:46 <warlord> no, you didn't ;)
22:11:18 *** twunder has quit IRC
22:11:31 *** twunder has joined #gnucash
22:55:03 <jsled> Wilddev: I re-wrote most of the SX code. Almost all of the not-gui-specific routines moved into not-gui-specific code, and I redid some of the UI ... the since-last-run dialog in particular.
22:55:26 <warlord> any plans to re-do the druid?
22:55:40 <jsled> The mortgage druid?
22:55:45 <warlord> yeah
22:55:49 <jsled> No.
22:56:02 <jsled> Not right now, but not longer-term either.
22:56:22 <warlord> minimally it should change how it handles escrow..
22:56:22 <jsled> The big things left are the FreqSpec -> Recurrence move and some unit tests.
22:57:10 <warlord> okay
22:57:15 <jsled> I've a couple of the trivial FreqSpec types reading into parallel Recurrences. And the way GncFrequency is setup is happily independent of either; so it can migrate over too.
22:58:19 <jsled> But I'm really itching to get this stuff basically sorted out for a 2.2, so I can look at the reports.
22:58:54 <Wilddev> jsled: nice
23:01:11 <jsled> warlord: Though another medium-term goal I have is to get the expression parser working off a real grammar, which would hopefully make easier and help support some of the things we've talked about in the past w.r.t. that druid.
23:01:36 <warlord> true.
23:01:54 <warlord> it would be REALLY nice to be able to handle P/I splits based off the actual account balance!
23:02:28 <jsled> aye. But my real focus is to make gnucash vastly simpler so other people can contribute that effectively. :)
23:03:01 <warlord> *nods*
23:04:43 <Wilddev> a worthy goal
23:09:34 <warlord> indeed.
23:10:07 <warlord> @tell esodan Sorry, I was out to dinner when you PM'd me. You can always have gncbot give me a message.
23:10:07 <gncbot> warlord: The operation succeeded.
23:11:14 <Wilddev> lol
23:26:05 *** dbr has joined #gnucash
23:39:38 *** twunder has quit IRC
23:46:37 *** warlord is now known as warlord-afk