2014-12-01 GnuCash IRC logs
00:56:56 *** MechtiIde has joined #gnucash
01:00:42 *** cartsoftware has joined #gnucash
01:07:45 *** cartsoftware has quit IRC
01:34:50 *** cartsoftware has joined #gnucash
01:38:58 *** O01eg has quit IRC
01:43:47 *** MechtiIde has quit IRC
01:47:07 *** jimvideo has quit IRC
01:49:42 *** cartsoftware1 has joined #gnucash
01:49:42 *** cartsoftware has quit IRC
01:55:12 *** Krzysiek_K has joined #gnucash
02:06:40 *** Krzysiek_K has left #gnucash
02:24:18 *** StuM has quit IRC
02:27:06 *** cartsoftware has joined #gnucash
02:31:50 *** cartsoftware2 has joined #gnucash
02:32:58 *** cartsoftware1 has quit IRC
02:39:25 *** cartsoftware has quit IRC
03:11:07 *** cartsoftware2 has quit IRC
03:23:48 *** wol has joined #gnucash
03:43:32 *** wol has quit IRC
04:02:32 *** wol has joined #gnucash
04:36:40 *** wol has quit IRC
05:10:40 *** GabrieleV_ has joined #gnucash
05:10:49 *** GabrieleV has quit IRC
05:10:50 *** GabrieleV_ is now known as GabrieleV
05:19:40 *** Gbarr has quit IRC
05:47:26 *** cartsoftware has joined #gnucash
06:02:59 *** Jimraehl1 has left #gnucash
06:05:23 *** Jimraehl1 has joined #gnucash
06:22:01 *** Infiltra1or is now known as Infiltrator
06:25:06 *** cartsoftware1 has joined #gnucash
06:31:38 *** cartsoftware has quit IRC
06:36:10 *** StuM has joined #gnucash
06:37:16 *** cartsoftware1 has quit IRC
06:55:46 *** GabrieleV_ has joined #gnucash
06:55:58 *** GabrieleV has quit IRC
06:56:00 *** GabrieleV_ is now known as GabrieleV
07:06:05 *** aqua___ has joined #gnucash
07:12:59 *** himaxx has joined #gnucash
07:29:32 *** andy has quit IRC
07:53:15 *** andy has joined #gnucash
07:54:23 *** wol has joined #gnucash
08:20:16 *** GabrieleV_ has joined #gnucash
08:20:17 *** GabrieleV has quit IRC
08:20:17 *** GabrieleV_ is now known as GabrieleV
08:28:44 *** wol has quit IRC
08:30:38 *** Gbarr has joined #gnucash
08:33:27 *** himaxx has joined #gnucash
08:34:24 *** wol has joined #gnucash
08:58:10 *** aqua___ has quit IRC
09:19:51 *** jralls has quit IRC
09:20:28 *** jralls has joined #gnucash
09:20:34 *** gncbot sets mode: +o jralls
09:24:35 *** cartsoftware has joined #gnucash
09:39:55 *** aqua___ has joined #gnucash
09:44:10 *** jralls has quit IRC
09:45:02 *** jralls has joined #gnucash
09:45:02 *** gncbot sets mode: +o jralls
09:48:07 *** ErKa has joined #gnucash
10:05:29 *** wol has quit IRC
10:06:38 *** aqua___ has quit IRC
10:14:32 *** StuM has quit IRC
10:17:36 *** O01eg has joined #gnucash
10:22:55 *** mlncn has joined #gnucash
10:25:00 *** wol has joined #gnucash
10:34:19 *** cartsoftware has quit IRC
10:42:18 *** wol has quit IRC
10:42:19 *** wol has joined #gnucash
10:44:09 *** lmat has joined #gnucash
10:44:53 *** smw has joined #gnucash
10:56:59 *** MechtiIde has joined #gnucash
10:58:38 *** fell has quit IRC
11:02:12 *** jralls has quit IRC
11:02:44 *** jralls has joined #gnucash
11:02:45 *** gncbot sets mode: +o jralls
11:08:21 <lmat> Does this colorful WM fella show up on IRC?
11:12:55 *** cartsoftware has joined #gnucash
11:15:51 <jralls> lmat: Not as far as I know. Why?
11:16:31 <lmat> jralls: Just curious if it was warlord ;-)
11:17:37 <warlord> uhhh.... no
11:20:22 *** Gbarr has quit IRC
11:20:45 <jralls> warlord: How about trimming the win32/build-logs directory on code? The list is getting pretty long. In particular all of the build-trunk and build-2.4 ones can go.
11:24:25 *** aqua___ has joined #gnucash
11:28:34 <warlord> jralls: sure
11:32:43 *** Gbarr has joined #gnucash
11:45:33 *** mlncn has quit IRC
11:48:49 *** GabrieleV_ has joined #gnucash
11:49:19 *** lmat has quit IRC
11:49:28 *** GabrieleV has quit IRC
11:49:29 *** GabrieleV_ is now known as GabrieleV
12:05:42 *** jimvideo has joined #gnucash
12:07:17 *** cartsoftware has quit IRC
12:08:20 *** cartsoftware has joined #gnucash
12:09:25 *** wol1 has joined #gnucash
12:09:47 *** Gbarr has quit IRC
12:10:22 *** Gbarr has joined #gnucash
12:12:11 *** mlncn has joined #gnucash
12:13:31 *** wol has quit IRC
12:14:32 *** mlncn-agaric has joined #gnucash
12:14:46 *** wol1 has quit IRC
12:14:47 *** jralls has quit IRC
12:15:39 *** jralls has joined #gnucash
12:15:39 *** gncbot sets mode: +o jralls
12:18:03 *** lmat has joined #gnucash
12:18:34 *** ErnstNeger has joined #gnucash
12:22:21 *** mlncn has quit IRC
12:22:21 *** lmat has quit IRC
12:24:55 *** mlncn-agaric has quit IRC
12:30:21 *** lmat has joined #gnucash
12:30:28 <lmat> @op
12:30:28 <gncbot> lmat: Error: You don't have the #gnucash,op capability. If you think that you should have this capability, be sure that you are identified before trying again. The 'whoami' command can tell you if you're identified.
12:39:02 *** Krzysiek_K has joined #gnucash
12:55:45 <gjanssens> jralls: great you found the cause of the failing jqplot charts
12:56:01 <gjanssens> I am unsure about the solution though
12:56:13 *** aqua___ has quit IRC
12:56:18 <gjanssens> It seems to me there is a / too much now on other platforms
12:56:42 <gjanssens> I mostly wonder what changed that it worked before and now it doesn't anymore
12:57:24 <gjanssens> First guess would be we used to pass a minw-ified path to that function while it now gets a real windows style path
12:58:53 <gjanssens> How did you manage to figure out that was the issue ?
13:06:22 *** cartsoftware has quit IRC
13:07:04 *** cartsoftware has joined #gnucash
13:13:47 *** cartsoftware has quit IRC
13:14:21 *** cartsoftware has joined #gnucash
13:29:40 <gjanssens> jralls: For comparison I reinstalled gnucash 2.6.3 which can generate the plots.
13:30:12 <gjanssens> The paths in there are also file://c:\program files\gnucash\... so with 2 slashes only
13:30:55 <gjanssens> Seems to me more recent webkits have a bug which fails to recognize c:\... as an absolute path :(
13:32:16 <gjanssens> Unless the webkit build system can/should be configured to do so (I haven't looked into this)
13:32:45 <gjanssens> On the other hand you have shown we can work around it by simply adding an additional /
13:34:10 <gjanssens> That is probably sufficient for now, although I prefer to see it special-cased for Windows and js include files only.
13:34:34 <gjanssens> Just to make it stand out as an exception.
13:35:01 * gjanssens wonders by the way whether css style sheets were also affected or not
13:35:30 <gjanssens> It may really be a bug in the javascript engine only.
13:37:57 <gjanssens> On the other issue with guile autocompile: do you have a more detailed error log for me ?
13:38:26 <gjanssens> I may have missed some paths to prepare the environment for autocompilation at build time
13:39:00 <gjanssens> The environment set up is similar to what we do during testing.
14:00:33 *** jralls has quit IRC
14:01:06 *** jralls has joined #gnucash
14:01:07 *** gncbot sets mode: +o jralls
14:08:47 <jralls> gjanssens: Actually, the file url is supposed to start with three slashes. Or more correctly, It starts with two, then the fqdn of the system with nothing being the default case, then the absolute path. See en.wikipedia.org/wiki/File_URI_scheme. Most browsers seem to be liberal about it, but WebKit 1.8 and 1.10 apparently don't.
14:11:51 <jralls> As for how I figured it out, I started debugging the processing of the web page and found that it was indeed getting to the javascript interpreter and that the message about the undefined variable $ was coming from the interpreter. That got me looking at what a "correct" jquery script looks like, and eventually found that the $(foo); is a jquery extension to javascript. At that point I realized that the problem must be tha
14:11:51 <jralls> t jquery wasn't getting loaded.
14:11:53 <gjanssens> jralls: ah, good to know
14:12:49 <gjanssens> Good hunting by the way
14:13:08 <jralls> Thanks. So the next issue for maint is how do we get guile to precompile on OSX?
14:13:23 <gjanssens> The wiki page leaves me with one question: is it possible to specify relative paths in uri-style ?
14:13:40 <gjanssens> Or is that simply not possible
14:14:34 <gjanssens> I'm asking because in our local file handling (xml://, sqlite:// and file://) I interpret uris with only two slashes are relative path files
14:14:58 <gjanssens> Which may be confusing against the official uri definition background
14:15:29 <gjanssens> But that's low priority now.
14:15:34 <jralls> AFAIK it's not legal to pass a relative path in a file: URI.
14:15:39 <gjanssens> Let's focus on the guile issue you're seeing
14:16:00 <gjanssens> During which part of the build did you get the error ?
14:16:24 <gjanssens> Can you pastebin the full error message in context ?
14:17:06 <jralls> Every place that guile precompiles and tries to load libgncmod-app-utils or libgncmod-gnome-utils.
14:18:41 <gjanssens> Meh, that's a lot :(
14:18:56 <gjanssens> And only those two ?
14:20:56 <jralls> I think only those two. I suspect that the reason is that they're built as .dylibs to allow them to be directly linked in test programs instead as loadable modules. In OSX either can be dlopened but only dylibs can be dynamically linked.
14:20:58 <gjanssens> jralls: where are the .dylib files stored at compile time (ie before make install is run)
14:21:40 *** MechtiIde has quit IRC
14:21:53 <jralls> In their respective .libs directories, like all of the other build products.
14:24:11 <jralls> IIRC there's a fiddle somewhere in GnuCash that tells guile how to load them, but the precompile isn't calling that.
14:25:56 <gjanssens> If the tests run fine on OS X it should be possible to use precompile as well
14:26:07 <gjanssens> I'm using the same script to set up the proper environment
14:26:19 <gjanssens> But perhaps I'm missing some paths
14:26:51 *** wizkid238 has quit IRC
14:27:02 *** wizkid238 has joined #gnucash
14:27:04 <gjanssens> Can you give me the dirs that fail to build ? That makes it easier for me to look in more detail .
14:27:43 <gjanssens> (It's not obvious from the makefiles which dirs actually use app-utils or gnome-utils)
14:28:45 <jralls> gjanssens: Pretty much all of them fail, but the first one is app-utils.
14:32:02 <jralls> The error message is ice-9/boot-9.scm:106:20: In procedure dynamic-link: file: "libgncmod-app-utils", message: "file not found". I don't think that the rest of the output is particularly enlightening.
14:32:32 <gjanssens> jralls: comparing src/app-utils/test/Makefile.am and src/app-utils/Makefile.am it looks like I'm missing the --gnc-module-dir entries
14:32:37 <gjanssens> That would make sense
14:33:02 <gjanssens> Can you try to add --gnc-module-dir ${top_builddir}/src/app-utils \
14:33:06 *** smw has quit IRC
14:33:26 <gjanssens> to GUILE_COMPILE_ENV in src/app-utils/Makefile.am and try again ?
14:34:44 <jralls> running...
14:35:59 <jralls> No joy.
14:37:04 <gjanssens> :(
14:40:52 <gjanssens> jralls: is the .dylib file already present when guile starts auto-compiling ?
14:41:17 <gjanssens> The build script waits for libgncmod-app-utils.la to be available
14:41:33 <gjanssens> I don't know if the dylib should be ready by then or not
14:47:55 <jralls> gjanssens: Yes, it's there. IIRC libtool writes the .la last after completing the link.
14:53:07 <jralls> gjanssens: Maybe there is something useful in the other output: The failure is in loading gettsxt, which calls (load-extension "libgncmod-app-utils" "scm_init_sw_app_utils_module")))
14:54:21 <jralls> It's still not obvious to me why that would work when running make check and not when running make.
14:54:29 *** cartsoftware has quit IRC
14:54:59 <gjanssens> gettsxt ? Never seen that before.
14:55:15 <gjanssens> Oh, another avenue to check: did you clear the autocompile cache ?
14:55:24 <gjanssens> I have seen issues when that was still around
14:56:22 <gjanssens> On linux it's in $HOME/.cache/guile/ccache/2.0-LE-8-2.0/....
14:58:14 <gjanssens> jralls: I figure you meant gettext...
14:58:31 <jralls> Yeah, sorry.
14:58:42 <gjanssens> That module gets loaded while loading app-utils.scm
14:59:10 <gjanssens> Note that app-utils.scm itself also calls load-extension libgncmod-app-utils
14:59:43 <gjanssens> Which didn't fail (it is done before gettext.scm is loaded)
15:00:01 <jralls> gjanssens: Removing the autocompile cache didn't change anything.
15:00:19 <gjanssens> Or perhaps I should ask: which module was it pre-compiling exactly ?
15:00:45 <gjanssens> app-utils.scm (which includes gettext.scm) ? or gettext.scm itself ?
15:01:05 <jralls> It says "3597: 15 [process-use-modules (((gnucash gettext)))]"
15:02:29 <jralls> But you meant "/Users/john/Development/Gnucash-Build/Gnucash-2.6-Leopard/inst/bin/guild compile -o business-prefs.go business-prefs.scm"
15:03:11 <jralls> In which (use-modules (gnucash gettext)) is the first line.
15:04:57 *** ErnstNeger has quit IRC
15:05:56 <gjanssens> Ah, that makes the case more interesting (give me a second)
15:08:09 <gjanssens> Hmm, what would happen if you replace that line with (use-modules (gnucash app-utils)) instead ?
15:09:34 <gjanssens> The other option would be to put the load-extension code in business-prefs as well before it loads gnucash gettext
15:10:46 <gjanssens> jralls: gnc-test-env export a variable DYLD_LIBRARY_PATH
15:11:14 <gjanssens> It that the typical environment variable that is used to locate .dylib files ?
15:11:24 <gjanssens> s/It/Is/
15:11:43 <jralls> gjanssens: DYLD_LIBRARY_PATH is for the dynamic linker, LD_LIBRARY_PATH is for dlopen.
15:12:18 <gjanssens> On guile irc someone suggests guile uses libltdl so we should set LTDL_LIBRARY_PATH
15:13:36 <gjanssens> I have my doubts though
15:14:12 <gjanssens> We don't set it for the installed gnucash either (or do you ?)
15:17:03 <jralls> No. And I just tested setting it without success.
15:22:54 <gjanssens> Ok then, did you also try to replace (use-modules (gettext)) with (use-modules (app-utils)) ?
15:23:26 <jralls> Maybe guild is doing something special with paths? When I tell the Makefile to use guile instead I get an error about on such file or directory .../src/app-utils/compile.
15:32:52 <jralls> Never mind that. It's because guild is really a two=part command and guile isn't.
15:35:50 <gjanssens> Ok
15:36:57 *** jralls_ has joined #gnucash
15:37:04 <gjanssens> I'm not sure the message "file not found" really relates to libgncmod-app-utils itself necessarily
15:37:35 <gjanssens> I seem to remember it's a pretty vague message that could equally mean that some of the libraries loaded by libgncmod-app-utils aren't found.
15:38:17 <gjanssens> I notice I'm missing several library lines compared to the makefile in the test directory
15:39:09 <jralls_> Yup. Give me a minute to copy them over and try again.
15:40:08 <gjanssens> Be sure to also keep the gnc-module-dir I was missing
15:41:20 <jralls_> That did it. Good thinking.
15:41:45 <gjanssens> Yay!
15:42:19 <gjanssens> Can I leave it to you to do the other failing ones as well ?
15:42:48 <jralls_> Sure.
15:42:54 <gjanssens> Tx
15:43:40 *** jralls has quit IRC
15:43:40 *** O01eg has quit IRC
15:43:40 *** Jimraehl1 has quit IRC
15:43:40 *** kim27 has quit IRC
15:43:40 *** kimmo2 has quit IRC
15:43:41 <jralls_> BTW, I tried again without the extra gnc-module-dir and it still worked.
15:43:48 *** jralls_ is now known as jralls
15:43:54 <jralls> @op
15:43:55 *** gncbot sets mode: +o jralls
15:45:20 *** kimmo2 has joined #gnucash
15:45:23 <jralls> gjanssens, warlord: New subject: Do either of you care about the build order on the VM? I think tag, maint, master would be better than the reverse, which is what is the order now.
15:45:44 *** O01eg has joined #gnucash
15:46:01 *** kim27 has joined #gnucash
15:46:22 <gjanssens> Thinking out loud: that tag build is initiated from the master setup
15:46:32 <gjanssens> Will that still work correctly if we build the tags first ?
15:47:53 <gjanssens> Looking at the build server scripts, I think it will still work
15:48:20 <gjanssens> The tag build script starts with updating the gnucash-on-windows directory just as the build-periodic script does
15:49:35 <gjanssens> On the other hand the tag build script will only fetch new tags in the gnucash directory, it won't pull new commits
15:49:51 <gjanssens> Will that also fetch tags on commits that are not pulled yet ?
15:50:20 <jralls> "pull" is just shorthand for "fetch && merge".
15:51:04 <gjanssens> Actually the command in the tags build script is 'git fetch -t'
15:51:28 <gjanssens> Will that fetch new commits and tags ?
15:51:38 <gjanssens> The man page is a bit vague on that for me
15:51:58 <gjanssens> If it does, I'm fine with swapping the order
15:53:14 <jralls> My reading of the man page is that it just gets tags. That's what it means when it says "the default refspecs are overridden and not used."
15:53:24 *** TommyT has joined #gnucash
15:54:33 <jralls> However, it will fetch the tagged commits and all of their predecessors, so the tagged build will still be correct.
15:55:05 <jralls> And then the master script will do the general pull.
15:56:54 <jralls> Hmm. Back to .go files: They're not in CLEANFILES. They should be, shouldn't they?
16:02:49 <gjanssens> Probably yes. A "make clean" doesn't remove them.
16:05:44 <jralls> Umm, yeah, that's why they need to be added to CLEANFILES.
16:07:01 <gjanssens> Makes sense except that it's confusing things like .o files do not have to be added...
16:07:12 * gjanssens never knows when it is needed and when not
16:08:14 <gjanssens> Re the tag builds: no further objections to swap them :)
16:08:26 <jralls> OK.
16:09:06 <gjanssens> It's bedtime over here
16:09:34 <gjanssens> Oh, before I forget:
16:09:48 <jralls> Make has default rules, run make -p | less to see them.
16:11:25 <gjanssens> https://bugzilla.gnome.org/show_bug.cgi?id=740934 may also be related to the guile-compile stuff
16:11:47 <gjanssens> I may not have tested the --enable-locale-specific-tax
16:11:53 <gjanssens> switch
16:13:31 * gjanssens is quickly trying to reproduce on fedora still...
16:13:35 <jralls> That switch bugs me. We don't use it on win32 or mac, either. Shouldn't it be enabled by default?
16:14:11 <jralls> Anyway, go on to bed. No need to fix it tonight.
16:15:09 <gjanssens> I don't know the motivation behind it.
16:15:35 <gjanssens> It enables some locale specific reports as far as I know
16:16:05 <jralls> It enables the German TXF report. It would enable others if we had them.
16:16:59 <gjanssens> No idea why that's not enabled by default. You could look at the commit that added it perhaps.
16:17:29 *** lmat has quit IRC
16:24:05 *** Krzysiek_K has left #gnucash
16:24:13 <gjanssens> The ugly thing about that hack is that even with --enable-locale-specific-tax set you'd only run into that bug if your locale is actually set to de_DE
16:24:27 <gjanssens> In all other cases the us tax report is loaded instead
16:24:56 <gjanssens> Anyway from glancing at the code the file txf-de_DE.scm needs the full app-utils dance
16:25:50 <gjanssens> It calls gnc-register-kvp-option-generator which is defined in app-utils, but that module isn't loaded
16:26:19 <gjanssens> (I probably ripped it out thinking it wasn't necessary since it compiled fine)
16:26:33 <gjanssens> But my locale in nl_BE, not de_DE
16:26:46 <gjanssens> Heh, quick analysis ;)
16:26:51 <gjanssens> Now I'm really off
16:26:56 <gjanssens> 'night
16:32:29 *** gjanssens has quit IRC
16:39:30 *** StuM has joined #gnucash
17:01:52 *** GabrieleV_ has joined #gnucash
17:02:02 *** GabrieleV has quit IRC
17:02:02 *** GabrieleV_ is now known as GabrieleV
17:17:34 *** cartsoftware has joined #gnucash
17:32:23 *** TommyT has quit IRC
17:46:48 *** himaxx has joined #gnucash
18:02:55 <jralls> warlord: I can't access code. Ping works, but neither SSH nor web access does. When I try to open an ssh terminal session it tels me "FATAL; errors found but logfile could not be created."
18:19:55 *** GabrieleV_ has joined #gnucash
18:19:56 *** GabrieleV has quit IRC
18:19:56 *** GabrieleV_ is now known as GabrieleV
18:24:18 *** jralls has quit IRC
18:25:00 *** jralls has joined #gnucash
18:25:01 *** gncbot sets mode: +o jralls
18:30:30 *** jralls has quit IRC
18:31:33 *** jralls has joined #gnucash
18:31:34 *** gncbot sets mode: +o jralls
19:14:04 *** jralls has quit IRC
19:14:56 *** jralls has joined #gnucash
19:14:57 *** gncbot sets mode: +o jralls
19:32:24 *** ErKa has quit IRC
19:46:06 <warlord> jralls: does it still do that?
19:50:37 <warlord> jralls: Looking at my mrtg graphs something happened at 1800 ET, but I can
19:50:41 <warlord> t figure out what it was.
19:54:30 <warlord> Logging into my router there is traffic... So not sure why mrtg isn't reporting it.
19:59:27 *** mlncn-agaric has joined #gnucash
20:01:00 *** lmat has joined #gnucash
20:28:01 <warlord> jralls: aha!!! There was an I/O error that caused the file system to crash.
20:28:51 <warlord> I'm rebooting code..
20:29:53 *** GabrieleV_ has joined #gnucash
20:31:02 *** GabrieleV has quit IRC
20:31:03 *** GabrieleV_ is now known as GabrieleV
20:32:28 *** [LasaK] has joined #gnucash
22:08:42 *** gncbot has joined #gnucash
22:08:52 *** warlord sets mode: +o gncbot
22:08:57 <warlord> jralls: okay, code should be back now.
22:20:01 <warlord> @op fell_
22:20:01 <gncbot> warlord: Error: You don't have the #gnucash,op capability. If you think that you should have this capability, be sure that you are identified before trying again. The 'whoami' command can tell you if you're identified.
22:20:18 <warlord> @op
22:20:19 *** gncbot sets mode: +o warlord
22:20:24 <warlord> @op fell_
22:20:24 *** gncbot sets mode: +o fell_
22:21:57 *** fell_ is now known as fell
22:22:34 <fell> Good Morning and thanks!
22:27:46 <warlord> you're welcome. Just testing.
22:34:58 *** puck has joined #gnucash
22:40:26 *** puck has quit IRC
22:41:06 *** puck has joined #gnucash
22:46:40 *** puck has quit IRC
22:47:16 *** puck has joined #gnucash
23:07:57 *** GabrieleV_ has joined #gnucash
23:08:36 *** GabrieleV has quit IRC
23:08:37 *** GabrieleV_ is now known as GabrieleV
23:31:14 <fell> warlord: wiki reports „145: Table './wiki/watchlist' is marked as crashed and should be repaired (localhost)“