Topic for #qi-hardware is now Copyleft hardware - http://qi-hardware.com | hardware hackers join here to discuss Ben NanoNote, atben / atusb 802.15.4 wireless, and other community driven hw projects | public logging at http://en.qi-hardware.com/irclogs
<DocScrutinizer>
anybody heard of solar storm that hits us just this moment?
<xiangfu>
kyak, so the correct is 'has FAILED' not 'was FAILED' ?
<kyak>
xiangfu: as far as my English grammar skills go, yes :)
<xiangfu>
ok.
<kyak>
DocScrutinizer: oh! and i thought i had a headache because of yesterday's party..
<DocScrutinizer>
hehe, same here, though actually I don't recall any party yesterday
<DocScrutinizer>
I just heard the storm is almost over and had no negative impact yet on COMSATs or power grids - but it will continue another few hours and may still get worse during that time
<DocScrutinizer>
the scary part in that sidenote in TV news was that this allegedly been known since 3 days, and I completely missed it
<kyak>
yeah, i actually heard it several times on TV, even though i watch TV only during breakfast :) it was like top news here
<DocScrutinizer>
and such an even at least potentially could cause EU wide blackouts
<DocScrutinizer>
event*
<DocScrutinizer>
which - for unclear reasons - makes me remember the fact that the sun twice a year(?) passes by exactly behind any arbitrary TV geostationary sat, which usually should cause massive interference in all dishes pointing at that sat and potentially even can fry LNBs
<DocScrutinizer>
funny coincidence: 1 year Fucku anniversary, nice date for an EU wide blackout
<DocScrutinizer>
which would also be a nice hard test for all nuclear power plants' powerless shutdown resilience
<wpwrak>
kyak: "was failed" would be the passive form. like in "several human rights activists were disappeared by the military regime"
wolfspraul has joined #qi-hardware
<DocScrutinizer>
LOL
<wpwrak>
kyak: so that would suggest some sort of sabotage that breaks our builds. now, combine this with the solar storm and you have a nice conspiracy story ;-)
<DocScrutinizer>
anyway could somebody kick that job outa the buildqueue please?
<DocScrutinizer>
seems it's triggered to rebuild and fail each time *anything* builds on the buildhost
<DocScrutinizer>
or is it that actually somebody adds this job over and over again, and never gets it right?
<kyak>
wpwrak: haha :)
<wpwrak>
DocScrutinizer: i think it's just some endless loop. maybe with some rate limit.
<kyak>
yeah, just an endless loop to build the whole image from scratch.. usually takes mor ethan one day
<kyak>
but if it fails early, it can become annoying :)
<DocScrutinizer>
does it make sense to compile that stuff again and again on an automated schedule, until eventually random bittoggles on HDD fix the errors?
<kyak>
it's supposed that someone will have a look and fix it ocasionally :)
<wolfspraul>
I fail to see the problem with that 1-line. just ignore the bot if you don't like it, but there are definitely people interested in regularly maintained and updated nanonote and milkymist software.
<wolfspraul>
maybe we can remove some redundancy?
<DocScrutinizer>
give me the URL and ssh credentials, I know the 2-letter cmd to fix it ;-)
<DocScrutinizer>
rm
<wolfspraul>
you don't use a nanonote and don't care about software updates for it
<wolfspraul>
that could explain your impatience with those log lines :-)
<kyak>
yeah, i'm also with the 1 line.. you can add the "Has FAILED" pattern to your IRC client, if you wish
<kyak>
s/also/also fine
<qi-bot>
kyak meant: "yeah, i'm also fine with the 1 line.. you can add the "Has FAILED" pattern to your IRC client, if you wish"
<DocScrutinizer>
I'm just concerned about the abstract waste of time and resources to build the obviously unchanged package over and over again
<wolfspraul>
that i agree with, if there is redundancy/noise, we should improve that
<DocScrutinizer>
that doesn't feel right
<kyak>
this is perhaps a good idea. the build bot should check if the repo has changed and it makes sense to rebuild
<wolfspraul>
but I think xiangfu is actively working on a new ben image right now, with testing, rebuilds, etc.
<DocScrutinizer>
indeed
<wolfspraul>
kyak: yes but I think that may already be the case
<wolfspraul>
but if not - yes, fully agree
<xiangfu>
kyak, (check repo) not in nanonote image build.
<kyak>
xiangfu: do you think it can be easily done, while you are on it? (check if the repo has changed since last FAILED build)
<xiangfu>
kyak, since the upstream package change every day.
<DocScrutinizer>
isn't it as simple as "rm /build-this" ?
<xiangfu>
kyak, that is easy. I already done that in milkymist stuff build (milkymist have 10 repos to check :)
<DocScrutinizer>
and ">build-this" when you edit something
<xiangfu>
DocScrutinizer, you mean disable/enable the build?
<DocScrutinizer>
or you actually use "find /$buildpackage -cnewer /$buildpackage/last-built"
<xiangfu>
( that isn't disabling the build once the compiler barfed up on it) yes. don't have that feature
<DocScrutinizer>
if find /$buildpackage -cnewer /$buildpackage/last-built; then build-it; else echo "nothing changed in $buildpackage since last build, skipping"; fi;
<xiangfu>
and it's hard to kill all process if the build already start. every time I have to manually to that by using htop
<DocScrutinizer>
if find /$buildpackage -cnewer /$buildpackage/last-built >/$buildpkg/changed-files; then echo "either those files changed: $(cat /$buildpkg/changed-files) or pkg never been built or compiler missed to create /$buildpkg/last-built tagfile. Building now"; build-it; else echo "nothing changed in $buildpackage since last build, skipping"; fi;
<xiangfu>
every build we want start from scratch.
<DocScrutinizer>
sure, it does
<DocScrutinizer>
depending on what you do in build-it
<DocScrutinizer>
it just doesn't start at all if there's already been a build that created the tagfile by "touch ./last-built" abd no file in the package changed since then
<DocScrutinizer>
obviously you want to touch ./last-built *after* the compiler created any new logfiles, *.o and whatnot else
<xiangfu>
what is the different on compare svn revision of openwrt packages repo?
<DocScrutinizer>
I.E. add "touch last-built" as very last line of your build-it job
<xiangfu>
if there is new commit rebuild from scratch. that is simple for me :-)
<DocScrutinizer>
meh
<DocScrutinizer>
quite obviously there's been no new commit for that openwrt pkg for the last few days
<DocScrutinizer>
yet it's rebuilt every day it seems
<DocScrutinizer>
the above sketched two shellscript lines would deal with that in a very convenient manner
<DocScrutinizer>
wait, you can make that one line, in one place:
<DocScrutinizer>
if find /$buildpackage -cnewer /$buildpackage/last-built >/$buildpkg/changed-files; then echo "either those files changed: $(cat /$buildpkg/changed-files) or pkg never been built or compiler missed to create /$buildpkg/last-built tagfile. Building now"; build-it; touch /$buildpkg/last-built; else echo "nothing changed in $buildpackage since last build, skipping"; fi;
<xiangfu>
DocScrutinizer, yes. that one have new commit check. but compile-openwrt-xburst.sh don't have and I don't have plan add it :-)
<xiangfu>
the step I work on nanonote is : 1. make sure the kernel compilef ine. 2. leave all other build to buildhost.
<DocScrutinizer>
well, I didn't even think of git, I just thought of some structure where a master script parses across a couple of $builpkg directories in something like /var/sources/all-packages/*
<DocScrutinizer>
#!/bin/sh\n #this is the master script that gets invoked once a day e.g. by cron\n \n cd /var/sources/all-packages; for buildpkg in *; do if find /$buildpackage -cnewer /$buildpackage/last-built >/$buildpkg/changed-files; then echo "either those files changed: $(cat /$buildpkg/changed-files) or pkg never been built or compiler missed to create /$buildpkg/last-built tagfile. Building now"; build-it; touch /$buildpkg/last-built; else
<DocScrutinizer>
echo "nothing changed in $buildpackage since last build, skipping"; fi; done
<DocScrutinizer>
umm
<DocScrutinizer>
#!/bin/sh\n #this is the master script that gets invoked once a day e.g. by cron\n \n cd /var/sources/all-packages; mkdir ../build-date; for buildpkg in *; do if find ./$buildpkg -cnewer ../build-date/$buildpackage >/tmp/${buildpkg}-changed-files; then echo "either those files changed: $(cat /tmp/${buildpkg}-changed-files) or pkg never been built or compile-job missed to create $(pwd)/../build-date/$buildpackage tagfile. Building
<DocScrutinizer>
now"; build-it $buildpkg; touch ../build-date/$buildpkg ; else echo "nothing changed in $buildpkg since last build, skipping"; fi; done
jekhor has joined #qi-hardware
<DocScrutinizer>
of course you have to check what find does when file referenced in -cnewer <file> doesn't exist
<DocScrutinizer>
o/, daywork (WinXP :-S )
<xiangfu>
if the $buildpackage mean openwrt package svn repo. that repo delete and checkout from scratch in every build.
<DocScrutinizer>
ouch
<xiangfu>
WinXP. there is Win 8 now. :-)
<DocScrutinizer>
well, that tells sth about confidence of big companies in that new crap from Redmond
<xiangfu>
DocScrutinizer, the openwrt make distclean , delete everything. include all package source(Makefiles)
<DocScrutinizer>
yeah, but only if HEAD != $whatever, AIUI your script. You could touch ../build-date/$packagename in that case
<DocScrutinizer>
anyway, my suggestion wasn't meant for git based infra. So nevermind
<xiangfu>
I have saved your script to my memo folder just in case :-)
<xiangfu>
wpwrak, you code on WPAN work out of box in linux v3.2.1.
<xiangfu>
wpwrak, no luck on ks7010. everytime update kernel. have to modify ks7010 driver.
<viric>
what is ks7010?
<wolfspraul>
viric: history :-)
<kristian1aul>
that one started tuesday?
<kristian1aul>
oops
kristianpaul has joined #qi-hardware
<wpwrak>
(wpan working out of the box) yippee! :)
liuqi has joined #qi-hardware
<kristianpaul>
wpwrak: testing last image build? :-)
liuqi has joined #qi-hardware
<wpwrak>
kristianpaul: just commenting on what xiangfu wrote
<wolfspraul>
we should officially drop ks7010 support and focus on ben-wpan, we just don't have the resources for ks7010 and it's a dead-end
<wolfspraul>
unless an avid ks7010 user speaks up...
<wolfspraul>
let me know, I'll read the backlog to face the ks7010 uprising :-)
<wolfspraul>
n8
<xiangfu>
my plan is will try 1 or 2 hours to fix the ks7010. if no luck. will stop working on that. keep test image. :-)
zrafa has joined #qi-hardware
emeb has joined #qi-hardware
GNUtoo has joined #qi-hardware
<qi-bot>
Jon Phillips: Via @qihardware qihardware: @milkymistvj <xiangfu> New video on DJ + Milkymist One : http://t.co/p9FmV9Yihttp://t.co/JbzOl2MY ( 178143337440747520@rejon - 5s ago via Ping.fm )