2017-10-03 GnuCash IRC logs
06:42:39 *** gncbot has joined #gnucash
06:42:46 *** ChanServ sets mode: +qo warlord warlord
06:42:49 *** warlord sets mode: +o gncbot
06:48:31 *** fabior has quit IRC
06:49:23 *** Jimraehl1 has joined #gnucash
07:00:21 *** phebus has quit IRC
07:02:27 *** phebus has joined #gnucash
07:12:42 *** jotrago has joined #gnucash
07:33:53 *** jotrago1 has joined #gnucash
07:34:24 *** jotrago has quit IRC
07:34:25 *** jotrago1 is now known as jotrago
07:39:02 *** jotrago has quit IRC
07:44:18 *** jotrago has joined #gnucash
07:51:52 *** jotrago1 has joined #gnucash
07:52:20 *** jotrago has quit IRC
07:52:21 *** jotrago1 is now known as jotrago
07:53:07 *** rickoehn has joined #gnucash
07:54:02 *** fabior has joined #gnucash
07:55:02 <gjanssens> fell: that probably means your previous push made it to code allright and then the server went down before finishing the push to github (you you got disconnected)
07:55:33 <gjanssens> Your second push today then didn't make a difference according to code, so no push to github was initiated
07:55:56 <gjanssens> Don't worry about it, the next push of a new commit will fix it automatically
08:01:37 *** gjanssens is now known as gjannsens
08:01:38 <gjannsens> .
08:01:38 <gncbot> gjannsens: Sent 7 weeks, 3 days, 18 hours, and 10 minutes ago: <fell> do you remember the glade msgctxt issue? I PR #170 the respective entries are missing.
08:08:35 <fell> OK, then I will look in the PR for a catalan website
08:14:14 <gjannsens> Good
08:16:10 *** gjannsens is now known as gjanssens
08:16:23 <gjanssens> fell: about the note that was still in the queue, it has a typo in my name. Did we already discuss it ? I was going through the logs, but didn't see it come up again
08:17:08 <fell> I believe, I added it to the PR
08:18:26 *** rickoehn has quit IRC
08:18:55 *** rickoehn has joined #gnucash
08:19:26 <fell> yep, commit 190a96f
08:19:28 <fell> Add missing glade msgctxt entries from intree build
08:19:29 <fell> 190a96f
08:19:32 *** rickoehn has left #gnucash
08:23:23 <gjanssens> You did indeed.
08:25:35 <gjanssens> I fear our cmake build is omitting these msgctxt entries as well, even when building out of tree.
08:29:18 *** fabior has quit IRC
08:30:33 *** gjanssens is now known as gjanssens_afk
08:32:34 *** rana has joined #gnucash
08:32:42 <rana> hi
08:32:42 <gncbot> rana: Sent 2 days, 10 hours, and 34 minutes ago: <warlord> If you want a response you have to stick around longer than a minute or two.
08:32:43 <gncbot> rana: Sent 2 days, 10 hours, and 34 minutes ago: <warlord> We haven't made any substantial changes to GnuCash's appearance in many years and see no reason to any time soon. That doesn't mean development is inactive; on the contrary it's quite active: https://github.com/Gnucash/gnucash/pulse/monthly
08:54:37 *** storyjesse has joined #gnucash
08:57:07 *** fabior has joined #gnucash
08:57:45 <warlord> rana: those are actually responses from jralls but gncbot wasn't around when he made them.
08:59:38 *** rana has quit IRC
09:06:01 <fell> Hm, How was the trick again to comment on single files of a commit?
09:06:20 *** gncbot sets mode: +o fell
09:06:28 <warlord> fell: what do you mean "comment on single files of a commit"?
09:06:43 <fell> at github
09:06:48 *** User has quit IRC
09:06:59 <fell> in the PRs
09:07:54 <warlord> Ah. I'm afraid I do not know that.. Sorry.
09:09:17 <fell> right above is the green button "Review changes" to comment on the whole commit, but in one tab one can comment on single files.
09:10:38 * warlord is not really a major github user. Maybe gjanssens_afk or jralls knows
09:10:58 <fell> NP
09:17:10 *** tuxd00d has joined #gnucash
09:22:53 *** tuxd00d has quit IRC
09:23:06 *** tuxd00d has joined #gnucash
09:25:46 <Chris> fell: you click on the commit, which displays the diff-patch, and hover over the code snippet which will bring out a '+' button to add a comment for a particular line
09:27:27 *** jotrago has quit IRC
09:35:36 *** jotrago has joined #gnucash
09:54:34 *** Chris has quit IRC
10:13:59 <fell> thx, Chris!
10:29:11 *** kael has joined #gnucash
10:35:16 *** kael has quit IRC
11:04:01 *** User has joined #gnucash
11:31:41 <jralls> fell: Weren't the msgctxt changes dependent upon a minimum version of gettext? I had to use an old distro (debian 8) to remake the tarballs for 2.6.18 because of a swig problem.
11:32:48 <jralls> warlord: Both you and gncbot are present. Have you resolved the network issue?
11:36:34 <fell> jralls, Mechtilde (debian stretch) had the same version as me. So it came out, you have to build out of tree to get them.
11:41:37 *** storyjesse has quit IRC
11:41:57 <jralls> fell: Stretch is Debian 9. I have a VM for it but it isn't behaving well enough yet to make a distribution. I'm also not sure about whether the swig it provides has the problem that F25 has with respect to guile-1.8. One wouldn't encounter that problem building for oneself.
11:44:13 <warlord> jralls: unfortunately no.
11:44:38 <warlord> still having major network issues here.
11:45:13 <fell> In https://wiki.gnucash.org/wiki/Translation#Background is a section beginning with AM_GNU_GETTEXT_VERSION(x.yy.zz)
11:45:34 <fell> there I collected the requirement
11:46:28 <fell> for glade we need 0.18.3
11:48:22 <fell> from July 2013
11:50:31 <fell> (1/2h break)
11:50:34 *** fell is now known as fell_afk
11:51:16 <warlord> jralls: if you have any suggestions for how to debug this further I'm all early. I've been running wireshark dumps between my edgerouter and the Arris modem to try to capture... But I'm just not seeing anything that stands out.
11:53:48 <jralls> warlord: IIUC yesterday's efforts you found that if you disconnected the VM host from the router the problem went away. Is that right? Is the switch in this case integrated with the router? You call it the "edge router"; does that mean that there are other routers inside of it?
11:54:36 <warlord> jralls, my architecture is Arris <-> Edgerouter <-> Switch <-> Everything else
11:54:45 <warlord> The vm server is sitting behind "everything else".
11:54:54 <warlord> (well, is part of "everything else")
11:55:02 <warlord> The edgerouter is literally a Ubiquity EdgeRouter
11:55:31 <warlord> It performs my NAT, GRE Tunnel (for my Class-C), and IPIP tunnel (for HEv6)
11:56:42 <warlord> Considering MOST of my upstream b/w is from the vm, it's not too surprising that disconnecting it clears up the network. But I'm not sure why I have to shut down everything..
11:57:31 <jralls> Maybe a bad network chip in the VM host?
11:57:54 <jralls> Do you have any network adapter PCI cards lying around?
11:58:19 <jralls> You have swapped out the cable already, right?
11:58:26 <warlord> The vm host has 2 ports (using LAG). I've tried both individually..
11:58:43 <jralls> They're probably supported by one chip.
11:59:17 <warlord> Could be. But that doesn't really explain how it can affect the RTT from the Arris
11:59:34 <warlord> It's almost as if it's causing some sort of buffering in the modem
12:03:38 <jralls> Badly formed MAC frames will cause all sorts of trouble and slow things down, but it does seem strange that it's on the link between the router and the modem. Each side of both the router and the switch should be electrically isolated.
12:04:44 <jralls> If you disconnect the router from the modem all of the ping times between hosts is normal?
12:06:15 *** jotrago1 has joined #gnucash
12:06:33 *** jotrago has quit IRC
12:06:34 *** jotrago1 is now known as jotrago
12:06:49 <warlord> I'm not seeing bad pings on my local network.
12:07:18 <warlord> Here's a snippet from MTR:
12:07:21 <warlord> My traceroute [v0.87]
12:07:21 <warlord> securerf.ihtfp.org (0.0.0.0) Tue Oct 3 12:07:01 2017
12:07:21 <warlord> Keys: Help Display mode Restart statistics Order of fields quit
12:07:21 <warlord> Packets Pings
12:07:21 <warlord> Host Loss% Snt Last Avg Best Wrst StDev
12:07:23 <warlord> 1. 192.168.248.1 4.1% 20692 0.9 0.8 0.6 124.2 7.6
12:07:26 <warlord> 2. 99.69.3.126 0.0% 20692 108.9 184.4 1.0 1455. 247.2
12:07:27 <warlord> 3. 107.223.192.1 0.1% 20692 812.7 456.9 2.4 8572. 518.5
12:07:29 <warlord> 4. 76.239.207.192 0.1% 20692 773.8 426.6 2.2 16225 725.4
12:08:02 <warlord> The first hop is fine, but the second hop (99.69.3.126) is my Arris modem. Notice the 108ms RTT, and the 812ms RTT to the next hop. Those should be under 5ms (and were until Sep 21)
12:11:18 <warlord> It's almost as if something is causing the Arris (or the Edgerouter) to buffer packets.
12:18:45 *** rana has joined #gnucash
12:19:09 <warlord> I'm not 100% convinced it's the VM server.. I was planning to pull it off the network again this afternoon to test it. The fact that this started around midnight just seems too coincidental for a random hardware problem or even a hacker attack. But I certainly didn't make any config changes, either.
12:19:38 <jralls> Hmm. If I ping 99.69.3.126 I get round-trip min/avg/max/stddev = 70.811/327.584/918.775/265.648 ms. 107.223.192.1 doesn't respond to ICMP from outside.
12:20:15 *** rana has quit IRC
12:20:41 <warlord> jralls: That's not surprising...
12:24:08 <warlord> if you mtr to that address I bet you see no loss along the path?
12:28:11 *** fell_afk is now known as fell
12:28:30 <warlord> (heading to lunch.. BIAB)
12:28:58 <fell> eating one after the other
12:28:59 <jralls> Don't have mtr... Have you tried http://netalyzr.icsi.berkeley.edu/?
12:44:36 <fell> Which gnu_gettext did you have on the VM, jralls
12:49:01 <jralls> fell: 0.19.3.
12:49:54 <fell> Then it should recognize the glade contexts
12:50:07 <fell> if built out of tree.
12:50:40 <fell> I suspect intltool to copy outdated stuff to /po
12:54:38 <jralls> Hmm, maybe gjanssens_afk is right that it's the cmake distcheck. That doesn't actually build anything until the check step. I'll experiment a bit. What's a filename to look for?
12:58:58 <fell> ../src/gnome-utils/gtkbuilder/gnc-frequency.glade.h:21,22,23
12:59:37 <fell> -23 +27,28
13:00:55 <gjanssens_afk> jralls: the files are there AFAIK, but any translatable message with message context (coming from glade files) is missing
13:01:05 <gjanssens_afk> You can search the po file for the msgctxt string.
13:01:12 <gjanssens_afk> (and leaving again...)
13:03:26 <warlord> jralls: I can't run that -- it requires java, which is disabled in modern firefox
13:07:39 <jralls> warlord: What about http://netalyzr.icsi.berkeley.edu/cli.html? ISTR you use Java...
13:24:01 <warlord> jralls: that might work.
13:34:29 <jralls> fell, gjanssens_afk: Confirmed that cmake isn't doing whatever it needs to include the msgctxt lines when it generates gnucash.pot. Note that autotools generates gnucash.pot during make dist but cmake does so during configuration.
13:46:49 <fell> I don't know Cmake, does it understand make-gnucash-potfiles.in?
13:49:22 <jralls> fell: It can be taught.
14:11:55 *** fabior has quit IRC
14:17:35 *** redarrow has joined #gnucash
14:19:14 <jralls> codemythe: Can you take a look at make-gnucash-potfiles.in and po/CMakeLists.txt and fix the latter to work like the former?
14:21:54 <fell> So the counterpart is in a different directory? Nice! ;-)
14:31:16 *** frakturfreak has joined #gnucash
14:36:15 <jralls> fell: That's mostly because of the somewhat weird way gettext does its autotools support... Makefile.in.in and such.
14:39:14 <fell> po/Makefile.in.in is forcecopyied by intltoolize and very outdated.
14:44:37 <jralls> fell: Maybe, but it's what generates the mo files.
14:51:13 <jralls> fell: But wait: make-gnucash-potfiles makes POTFILES.in. It's not what makes gnucash.pot from them. Makefiles.in.in does, calling intltool-extract.
14:52:00 <jralls> I think gjanssens_afk has already checked more than once that cmake and autotools produce the same results for POTFILES.in.
15:00:50 <jralls> Sorry, passing intltool-extract and xgettext to intltool-update --package gnucash --pot. Same command in both CMakeLists.txt and Makefile.in.in. So whatever is different is buried in there somewhere.
15:01:49 <jralls> I'm summoned for lunch.
15:01:54 *** jralls is now known as jralls_afk
15:07:09 *** tuxd00d has quit IRC
15:19:07 *** jotrago1 has joined #gnucash
15:20:01 *** jotrago has quit IRC
15:20:03 *** jotrago1 is now known as jotrago
15:46:57 *** jquin has joined #gnucash
15:56:38 *** User has quit IRC
16:00:10 *** gjanssens_afk is now known as gjanssens
16:01:09 <warlord> jralls_afk, gjanssens, fell --- I'd like to unplug the VM system again for about an hour to test my networking. is that okay with you?
16:01:22 <gjanssens> jralls_afk, fell: POTFILES.in is fine indeed. It's the same wether generated by autotools or cmake
16:02:28 <gjanssens> jralls_afk: as you conclude there must be a subtle difference in the way intltool-extract/intltool-update is used that is responsible for the difference.
16:02:35 <gjanssens> warlord: that's ok for me.
16:08:44 <warlord> gjanssens: thanks. fell, okay with you?
16:14:59 <warlord> fell...??
16:15:49 <warlord> going...
16:15:57 <warlord> going...
16:16:00 <warlord> gone..
17:33:43 *** gncbot has joined #gnucash
17:35:08 <gjanssens> The one that caught my attention was /mingw32/etc/fonts so I tried to add this manually to the bundle-installed gnucash, and now gnucash doesn't crash any more...
17:35:18 <gjanssens> jralls: looks like that's what was missing
17:35:33 <jralls> Yay!
17:36:22 <gjanssens> Yay indeed :)
17:38:09 <gjanssens> I got my clue from the old mingw based installer. I still had one such installation on my XP box and decided to check what else was in there.
17:38:28 <gjanssens> Anyway, I think I'll leave you now. Time to get some rest.