04:50 <Joerg-Neo900> http://laforge.gnumonks.org/blog/20160920-openmoko_10years/

08:18 <wpwrak> (om) oh, i'm sure that this opportunity will come back. e.g., when some "locky" for android 4.x appears. then all the people with devices abandoned by their manufacturers will have an extremely rude awakening.

08:21 <wpwrak> (abandoned) and of course, without any alternative choice, thanks to the design choices of that manufacturer

08:23 <wpwrak> or maybe it won't be locky but mirai. cellular ddos could be fun, too :)

13:45 <ceene> wpwrak: eeshow does in fact segfault when pressing key up/down

13:45 <ceene> i run it with make view from ee/hw

13:45 <ceene> i've compiled eeshow from git version ahd same thing happens

13:47 <ceene> #1 0x0000000000419260 in hover_history (user=0x112a400, on=false, dx=0, dy=0) at gui/history.c:304

13:47 <ceene> 304s = vcs_git_summary_for_pango(h->vcs_hist, fmt_pango);

13:47 <ceene> (gdb) print h->vcs_hist

13:47 <ceene> $3 = (struct vcs_hist *) 0x0

14:16 <wpwrak> hmm. works fine here.

14:17 <wpwrak> ceene: but you said you also donn't get the thingies in the upper left corner ?

14:32 <wpwrak> if you don't get anything up there, that would point to a more severe underlying problem

14:32 <wpwrak> i.e., eeshow not being able to get the history

14:33 <ceene> that's right, yes

14:34 <wpwrak> what does it print when starting ? i get this: https://neo900.org/stuff/paste/ooHaix6z.txt

14:35 <ceene> only the first ignoring "maxheight" line

14:35 <ceene> related_other_repo doesn't appear

14:35 <wpwrak> did you copy the files out of the repo directory ?

14:35 <ceene> nop, i'm under the repo

14:36 <ceene> i can git log, git commit and whatever

14:38 <wpwrak> how about this ? invoking eeshow directly, with debugging output

14:39 <ceene> how?

14:39 <wpwrak> first line, eeshow -v -v -v -N 1 neo900.pro

14:40 <ceene> not much

14:41 <ceene> http://paste.debian.net/899509/

14:42 <wpwrak> hmm. blissfully unaware of git

14:49 <wpwrak> fixed the crash on Up/Down if there's no history.

14:49 <wpwrak> now ... why does eeshow think there is none if there clearly is ...

16:13 <Joerg-Neo900> libgit version?

18:06 <Joerg-Neo900> protect your Port 7547

18:17 <Joerg-Neo900> and disable TR-069 if any possible!

18:19 <Joerg-Neo900> ceene: could you send the pubkey?

22:15 <wpwrak> ceene: i added a bit more diagnostics to the function that should detect that neo900.pro is in a git repo, but apparently fails to do so. maybe that will provide a hint. kinda mysterious, though.

22:24 <Joerg-Neo900> wpwrak: did you consider version problems with that libgit-$wahetever?

22:25 <Joerg-Neo900> wasn't there this weird "symbol _init_ not defined" or whatever, when I built eeshow?

22:26 <Joerg-Neo900> I seem to very cloudy recall a comment by you that started with words like "let's hope this..."

22:26 <Joerg-Neo900> I might be wrong

22:27 <wpwrak> i rebuilt eeshow on my (new) ubuntu 16.10, and it still works as expected. so it doesn't look like any recent incompatibility.

22:27 <wpwrak> (let's hope) well, there ought to be many of these ;)

22:33 <wpwrak> i don't have any specific suspicion of what could be wrong at the moment. it seems that vcs_git_try fails to detect that neo900.pro is in a git repo, for an unknown reason. maybe the new diagnostics will reveal more.

22:35 <Joerg-Neo900> maybe a nice approach to gather some helpful background info was to issue a few git ${whatever:-status}|pastebinit commands on ceene's system?

22:37 <Joerg-Neo900> also `echo $PWD >>analysis.log`

22:38 <Joerg-Neo900> also `ls -laR >>analysis.log`

22:38 <Joerg-Neo900> I'm the git noob so I can't help much here :-)

22:39 <Joerg-Neo900> cygwin? :-o

22:40 <Joerg-Neo900> whatever, I guess we won't see ceene around before Wednesday ;-)

22:40 <wpwrak> ceene already mentioned that the usual git operations work. what the diagnostics obtained so far indicate is that eeshow doesn't even try to use this like a git repo. so it seems that the detection in vcs_git_try fails. but i don't know why that would happen. if the new diagnostics don't reveal anything, we can try some things with eetest, which is more low-level.

22:41 <Joerg-Neo900> could gdb enable a strace alike tracing for the lib only?

22:42 <Joerg-Neo900> I *guess* in the end all git operations are just file open() read() write() functions?

22:43 <Joerg-Neo900> stat()

22:43 <wpwrak> hmm, not really. with gdb, you better know what you're looking for. but eetest (perhaps with a new mode for debugging this problem) would be simple enough to allow for using strace directly

22:43 <Joerg-Neo900> k

22:43 <Joerg-Neo900> :-)

22:43 <wpwrak> the "not really" was about gdb. yes, git is just file operations. no hidden mysql nightmare or such ;-)

22:44 <Joerg-Neo900> got it that way

22:49 <Joerg-Neo900> ARRRRGH, eeshow in itself doesn't have the version linit (yet?)

22:49 <Joerg-Neo900> limit

22:50 <Joerg-Neo900> happily ran away busy, grabbing 2GB before I was able to kill it ;-P

22:51 <Joerg-Neo900> it *really* should assume a default of 20 for -N

22:53 <Joerg-Neo900> echo "(unlimited if omitted or 0)" | sed 's/omitted or//; s/.*\)$/\&. default: 20\)/'

23:05 <Joerg-Neo900> FWIW:

23:05 <Joerg-Neo900> jr@saturn:~/Dokumente/Neo900/projbin/neo900-proj/ee/hw> strace -e trace=file eeshow -N 20 neo900.pro 2>&1| grep '\.git'|head -n 2000|pastebinit

23:05 <Joerg-Neo900> http://susepaste.org/20676987

23:07 <Joerg-Neo900> ((should assume a default of 20 for -N)) I guess that's a very simple "one keypress: 2" patch, plus ficing the --help output

23:08 <Joerg-Neo900> fixing*

23:16 <Joerg-Neo900> might be useful for finding where things start going haywire, when you use it for a diff, after cleaning up absulte pathnames to relative ones by sed

23:16 <Joerg-Neo900> absolute*

23:17 <Joerg-Neo900> lemme already do this

23:20 <Joerg-Neo900> jr@saturn:~/Dokumente/Neo900/projbin/neo900-proj/ee/hw> strace -e trace=file eeshow -N 20 neo900.pro 2>&1| grep '\.git'|head -n 2000|sed 's&/home/jr/Dokumente/Neo900/projbin/neo900-proj/ee/&+#+/&g'|pastebinit

23:20 <Joerg-Neo900> http://susepaste.org/85985175

23:24 <Joerg-Neo900> and finally the canonical version:

23:24 <Joerg-Neo900> ee/hw> strace -e trace=file eeshow -N 20 neo900.pro 2>&1| grep '\.git'|head -n 2000|sed "s&$(dirname $(pwd))&+#+/&g"|pastebinit

23:24 <Joerg-Neo900> http://susepaste.org/12974517

23:31 <Joerg-Neo900> wpwrak: please let me know if you think more lines or other parameters fro grep or strace might be needed

23:37 <wpwrak> too complex :) and you're suppressing information that may be useful. a better test would be more something like strace -o log /path/to/eetest -C neo900.pro

23:37 <wpwrak> err, strace -o log /path/to/eetest -C HEAD:neo900.pro

23:38 <Joerg-Neo900> no idea about eetest

23:38 <wpwrak> but first test would be to see if eetest -C HEAD:neo900.pro works. if it does, we need a new option in eetest, for a more specific tets. if it doesn't we have our smoking gun.

23:39 <wpwrak> eetest is part of eeshow, but doesn't get installed (by make install). it's only for debugging, doesn't do anything actually useful for normal use.

23:39 <Joerg-Neo900> may we do a little bet that the problem shows up in first 80 lines of http://susepaste.org/12974517 ? ;-)

23:39 <wpwrak> well, unless you like eetest -H . ;-)

23:41 <Joerg-Neo900> tbh I strongly suspect a very simple pathname issue, maybe around CWD or somesuch

23:41 <jonsger> http://laforge.gnumonks.org/blog/20160920-openmoko_10years/ quite interesting, but I kind of outdated Fairphone uses 100MB blob not 70MB :(

23:41 <wpwrak> (eetest -H shows the history and dumps it as text. a bit like tig and such)

23:44 <wpwrak> Joerg-Neo900: yes, could be a pathname thing. that has happened before. but those problems usually occur a little bit later. this one seems to go wrong very early, with eeshow hardly having done anything yet. we'll see.

23:44 <wpwrak> the new diagostics shold also show the path name if it fails where it seems it fails

23:44 <Joerg-Neo900> huh?

23:45 <Joerg-Neo900> aiui it simply doesn't show any git history

23:46 <Joerg-Neo900> and my strace above starts as soon as (AIUI) the first git related file access happens

23:47 <Joerg-Neo900> if it accesses 2000 times the git repo before anything goes haywire so it doesn't show a git history at all, then I dunno

23:48 <Joerg-Neo900> I said "within first 80 lines", I wast temped to day within first 10

23:48 <Joerg-Neo900> I said "within first 80 lines", I was temped to say within first 10

23:49 <Joerg-Neo900> did you actually look at it?

23:51 <Joerg-Neo900> if ceene's log would differ from first line already, then we know it's something way earlier

23:53 <Joerg-Neo900> tbh very first thing i'd look into would be a ls -la ../*

23:56 <Joerg-Neo900> ls -la ../.git .git