00:38
gric has joined #m-labs
02:45
evilspirit has joined #m-labs
03:23
fengling has joined #m-labs
04:30
fengling has quit [Quit: WeeChat 1.4]
04:38
evilspirit has quit [Ping timeout: 240 seconds]
04:42
fengling has joined #m-labs
05:01
fengling has quit [Quit: WeeChat 1.4]
05:04
fengling has joined #m-labs
05:38
evilspirit has joined #m-labs
05:48
<
whitequark >
sb0: no, you're confused about the role of qemu
05:48
<
whitequark >
qemu provides a device model here
05:48
<
whitequark >
whereas xen provides virtualized code execution
05:49
<
whitequark >
since compiling is a CPU-bound task, this pays off
05:49
<
whitequark >
if we had a disk-bound task, there would be little difference, for example
05:50
<
whitequark >
as for connecting to X, I don't think you can do that with xen, no; the qemu device model binary doesn't have any X in it
05:51
<
whitequark >
rjo: ok. yes. I see.
05:51
<
whitequark >
we need another build machine, this time it will be an ubuntu-trusty64. it will run the conda-lin64 buildbot.
06:24
kyak_ is now known as kyak
06:24
fengling has quit [Ping timeout: 240 seconds]
06:33
fengling has joined #m-labs
07:34
<
sb0 >
qemu-system-i386
*is* loading X and SDL libs. and I do remember using qemu-system-lm32 that popped out a X11 windows.
07:38
<
whitequark >
hm, hang on, I have a better idea
07:38
<
whitequark >
why don't you install tightvncserver *on the windows machine*?
07:39
<
sb0 >
that would solve the current bugs, but also I'm still trying to understand why you wrote a vnc client
07:39
<
whitequark >
as far as I'm aware you can't use sdl or x from under xen
07:40
<
sb0 >
(and assuming, of course, that tightvncserver isn't another clusterfuck like the cygwin sshd)
07:41
<
sb0 >
why do we need two build machines?
07:42
<
whitequark >
because host's glibc is too new
07:42
<
whitequark >
well, three
07:42
<
whitequark >
I don't mean a physical machine, we just need a trusty64 VM
07:43
<
sb0 >
ok. and probably more RAM as well
07:43
<
whitequark >
more RAM never hurts
07:44
<
sb0 >
right now there are two windows VMs running and that would make three VMs, plus it needs to run ISE, right?
07:44
<
whitequark >
two ISE copies at once, right now
07:44
<
sb0 >
isn't it swapping?
07:45
<
whitequark >
pretty sure it does
07:45
<
whitequark >
note that VMs cannot be swapped out, I think
07:45
<
whitequark >
at least Xen refuses to allocate a VM with more RAM than physical
07:46
<
whitequark >
yeah, right, the idea is that VMs should do their own swap management
07:46
<
whitequark >
since Xen's would be inefficient
07:47
<
whitequark >
this is much less of a problem with Linux VMs since they can use balloon (dynamically request RAM from Xen)
07:48
<
sb0 >
anyway, more RAM is cheap and in theory easy
07:49
<
whitequark >
there can be up to 16GB of RAM on this motherboard
07:49
<
whitequark >
that's the first time I hear about SDL. yeah, I guess a VNC client wasn't necessary then.
07:54
<
whitequark >
hm, so right now buildserver has 8GB of RAM
07:54
<
whitequark >
around 3GB is for dom0, the rest is allocated to VMs
07:54
<
whitequark >
so we at least have room for an ubuntu64 vm
08:14
<
sb0 >
would have been surprising if there was no reliable graphical display out of VMs. that's one of the essential features of a widely used program
08:16
<
sb0 >
cr1901_modern, zooming is much better, but still a bit fof
08:17
<
sb0 >
cr1901_modern, I think you really send mouse wheel events from the sliders into the axis (or otherwise make the slider area also zoom reactive)
08:19
<
sb0 >
cr1901_modern, when I zoom in a lot between the cursors, the cursors do not disappear (as they become out of range) - they should. but i guess you are already planning to implement that?
08:19
<
sb0 >
better if they disappear progressively, i.e. the cursor can be straddling the widget borders. that will be needed for nice scrolling too.
08:33
sb0_ has joined #m-labs
08:35
sb0 has quit [Ping timeout: 248 seconds]
09:05
fengling has quit [Ping timeout: 240 seconds]
09:09
<
whitequark >
the webkit people are insane
09:10
fengling has joined #m-labs
09:15
<
larsc >
what did they do now?
09:18
<
whitequark >
you need python, ruby and perl to build qtwebkit
09:32
<
whitequark >
seems like i have a decent qt package
09:32
<
whitequark >
now to make it work.
09:33
<
whitequark >
bb-m-labs: force build --props=package=qt5 conda-win64
09:33
<
bb-m-labs >
build forced [ETA 37m18s]
09:33
<
bb-m-labs >
I'll give a shout when the build finishes
09:33
<
whitequark >
good start
09:33
<
whitequark >
bb-m-labs: force build --props=package=qt5 conda-win64
09:33
<
bb-m-labs >
build forced [ETA 37m18s]
09:33
<
bb-m-labs >
I'll give a shout when the build finishes
09:33
<
GitHub38 >
conda-recipes/master 18fdd3b whitequark: qt: remove.
09:33
<
GitHub38 >
conda-recipes/master 641aa59 whitequark: qt5: add.
09:38
rohitksingh has joined #m-labs
09:44
<
whitequark >
bb-m-labs: force build --props=package=qt5 conda-lin64
09:44
<
bb-m-labs >
build forced [ETA 43 seconds]
09:44
<
bb-m-labs >
I'll give a shout when the build finishes
09:56
<
whitequark >
who writes batch files like that
09:56
<
whitequark >
aaaargh
10:03
<
whitequark >
guess what, it didn't actually build on windows
10:03
<
GitHub140 >
conda-recipes/master 80281d4 whitequark: qt5: add.
10:03
<
whitequark >
because it didn't do if errorlevel == 1 exit 1
10:03
<
whitequark >
(the equivalent of set -e)
10:04
<
whitequark >
cretins
10:04
<
whitequark >
bb-m-labs: force build --props=package=qt5 conda-lin64
10:04
<
bb-m-labs >
build forced [ETA 43 seconds]
10:04
<
bb-m-labs >
I'll give a shout when the build finishes
10:04
<
whitequark >
bb-m-labs: force build --props=package=qt5 conda-win64
10:04
<
bb-m-labs >
build forced [ETA 26m30s]
10:04
<
bb-m-labs >
I'll give a shout when the build finishes
10:18
<
whitequark >
bb-m-labs: force build --props=package=qt5 conda-lin64
10:18
<
bb-m-labs >
build forced [ETA 43 seconds]
10:18
<
bb-m-labs >
I'll give a shout when the build finishes
10:18
<
GitHub115 >
conda-recipes/master fd7e145 whitequark: qt5: add.
10:19
<
whitequark >
bb-m-labs: force build --props=package=qt5 conda-lin64
10:19
<
bb-m-labs >
build forced [ETA 43 seconds]
10:19
<
bb-m-labs >
I'll give a shout when the build finishes
10:19
<
GitHub19 >
conda-recipes/master 490f90c whitequark: qt5: add.
10:25
<
whitequark >
bb-m-labs: force build --props=package=qt5 conda-lin64
10:25
<
bb-m-labs >
build forced [ETA 43 seconds]
10:25
<
bb-m-labs >
I'll give a shout when the build finishes
10:25
<
GitHub97 >
conda-recipes/master 2a9ff56 whitequark: qt5: add.
10:31
<
whitequark >
bb-m-labs: force build --props=package=qt5 conda-lin64
10:31
<
bb-m-labs >
build forced [ETA 43 seconds]
10:31
<
bb-m-labs >
I'll give a shout when the build finishes
10:33
fengling has quit [Quit: WeeChat 1.4]
10:34
<
GitHub85 >
conda-recipes/master 33043ca whitequark: qt5: add.
10:34
<
whitequark >
bb-m-labs: force build --props=package=qt5 conda-lin64
10:34
<
bb-m-labs >
build forced [ETA 43 seconds]
10:34
<
bb-m-labs >
I'll give a shout when the build finishes
10:34
<
whitequark >
bb-m-labs: force build --props=package=qt5 conda-win64
10:34
<
bb-m-labs >
build forced [ETA 26m30s]
10:34
<
bb-m-labs >
I'll give a shout when the build finishes
10:34
<
GitHub175 >
conda-recipes/master 2430243 whitequark: qt5: add.
10:51
<
whitequark >
hm, now it can't find ICU
10:51
<
whitequark >
but I gave it ICU
10:52
<
whitequark >
/var/lib/buildbot/slaves/debian-stretch-amd64-1/miniconda/envs/_build/lib
10:52
<
whitequark >
g++ -Wl,--gc-sections -Wl,-O1 -Wl,-rpath,/var/lib/buildbot/slaves/debian-stretch-amd64-1/miniconda/envs/_build/lib/qt5 -o ../../../bin/qlalr .obj/compress.o .obj/cppgenerator.o .obj/dotgraph.o .obj/lalr.o .obj/main.o .obj/parsetable.o .obj/recognizer.o .obj/grammar.o -L/var/lib/buildbot/slaves/debian-stretch-amd64-1/miniconda/envs/_build/lib -L/var/lib/buildbot/slaves/debian-stretch-amd64-1/miniconda/conda-bld/work/qt-everywhere-opensource-src-
10:52
<
whitequark >
/usr/bin/ld: warning: libicui18n.so.56, needed by /var/lib/buildbot/slaves/debian-stretch-amd64-1/miniconda/conda-bld/work/qt-everywhere-opensource-src-5.5.1/qtbase/lib/libQt5Core.so, not found (try using -rpath or -rpath-link)
10:52
<
whitequark >
$ ls $(pwd)/libicui18n.so.56/var/lib/buildbot/slaves/debian-stretch-amd64-1/miniconda/envs/_build/lib/libicui18n.so.56
10:52
<
whitequark >
what the hell is your problem exactly, gcc
10:53
<
sb0_ >
can you ldd that $(pwd)/libicui18n.so.56/var/lib/buildbot/slaves/debian-stretch-amd64-1/miniconda/envs/_build/lib/libicui18n.so.56 ?
10:54
<
whitequark >
it's fine of course
10:55
<
whitequark >
what the fuck
10:56
<
whitequark >
it tries all the system paths and then /var/lib/buildbot/slaves/debian-stretch-amd64-1/miniconda/envs/_build/lib/qt5/libicuuc.so.56
10:56
<
whitequark >
and completely ignores -L/var/lib/buildbot/slaves/debian-stretch-amd64-1/miniconda/envs/_build/lib
10:56
<
sb0_ >
.so? isn't gcc looking for .a?
10:57
<
whitequark >
why would it look for .a?
10:57
<
whitequark >
it's a dependency of Qt5Core
10:57
<
whitequark >
OH that's why
10:57
<
whitequark >
.... hm.
10:58
<
whitequark >
right, it's not really designed to look outside of the prefix libdir
10:58
<
whitequark >
and the prefix libdir is lib/qt5.
10:58
<
whitequark >
that... makes a certain amount of sense
10:58
<
whitequark >
there was some obscure linker flag.
10:59
<
whitequark >
yeah, I could just add a second rpath, why the hell not
11:00
<
whitequark >
there's even an option for that in qt's configure, cool
11:00
<
GitHub187 >
conda-recipes/master 36f6aad whitequark: qt5: add.
11:01
<
whitequark >
bb-m-labs: force build --props=package=qt5 conda-lin64
11:01
<
bb-m-labs >
build forced [ETA 43 seconds]
11:01
<
bb-m-labs >
I'll give a shout when the build finishes
11:01
<
whitequark >
hm, cannot find system freetype on windows
11:02
<
whitequark >
ah, another problem in the original recipe
11:03
<
whitequark >
I should've just rm'd it instead of taking it as a base.
11:03
<
GitHub42 >
conda-recipes/master af4a924 whitequark: qt5: add.
11:03
<
whitequark >
bb-m-labs: force build --props=package=qt5 conda-win64
11:03
<
bb-m-labs >
build forced [ETA 26m30s]
11:03
<
bb-m-labs >
I'll give a shout when the build finishes
11:03
<
sb0_ >
original recipe == the spyder qt5?
11:03
<
whitequark >
and it's also in conda-recipes/
11:03
<
whitequark >
on github
11:03
<
sb0_ >
well they're not actually using it...
11:03
<
whitequark >
it's phenomenally crappy
11:03
<
whitequark >
yes, that explains why
11:21
acathla` has joined #m-labs
11:21
acathla has quit [Read error: Connection reset by peer]
11:21
acathla` is now known as acathla
11:21
acathla has joined #m-labs
11:21
acathla has quit [Changing host]
11:23
<
whitequark >
hahahahaha wow
11:23
<
whitequark >
I knew NTFS is bad
11:23
<
whitequark >
but our Windows VM spends TWENTY SIX MINUTES deleting old Qt and unpacking new
11:23
<
whitequark >
Qt has... 153 thousand files
11:24
<
whitequark >
and NTFS is really bad with lots of small files.
11:31
acathla` has joined #m-labs
11:32
acathla has quit [Read error: Connection reset by peer]
11:34
acathla` is now known as acathla
11:34
acathla has quit [Changing host]
11:34
acathla has joined #m-labs
11:42
<
whitequark >
it worked. amaze. wow.
11:42
<
whitequark >
sb0_: can you test the qt package?
11:43
<
whitequark >
well, it's called "qt5" again
11:43
<
whitequark >
oh, no pyqt...
11:44
<
sb0_ >
whitequark, yes. can you update the artiq package and test it on linux + windows?
11:44
<
whitequark >
no, I can not.
11:44
<
whitequark >
I don't have a working conda install anywhere I have X.
11:45
<
sb0_ >
then that's a good opportunity to verify that the full installation instructions in the manual still work
11:45
<
whitequark >
what do you mean?
11:46
<
whitequark >
if I do the full installation, I'll just use the debian qt5 package
11:46
<
sb0_ >
that when you take a fresh anaconda or miniconda, and input the exact command in the manual, you get a working artiq.
11:46
<
sb0_ >
someone needs to test this
11:46
<
whitequark >
no, you don't understand
11:46
<
whitequark >
conda doesn't work on my system
*at all*
11:46
<
rjo >
whitequark: a trusty chroot would be fine.
11:51
<
sb0_ >
whitequark, conda works on my system, i can reinstall it and try out the artiq install...
11:51
<
whitequark >
sb0_: not yet, since pyqt is not built yet
11:51
<
whitequark >
let me fix it up
11:53
<
GitHub15 >
conda-recipes/master 0530c3b whitequark: pyqt: rename to pyqt5 again.
11:53
<
whitequark >
bb-m-labs: force build --props=package=pyqt5 conda-lin64
11:53
<
bb-m-labs >
build forced [ETA 20m13s]
11:53
<
bb-m-labs >
I'll give a shout when the build finishes
11:53
<
whitequark >
sb0_: does anaconda ship quamash too?
11:53
<
whitequark >
should we name it quamash5?..
11:54
<
sb0_ >
isn't quamash qt version agnostic (and noarch)?
11:54
<
whitequark >
it has a dep on qt>=5
11:54
<
whitequark >
pyqt>=5 even
11:55
<
sb0_ >
can the conda dependency system express pyqt | pyqt5?
11:55
<
whitequark >
no clue
11:56
<
whitequark >
bb-m-labs: force build --props=package=pyqt5 conda-lin64
11:56
<
bb-m-labs >
build forced [ETA 20m13s]
11:56
<
bb-m-labs >
I'll give a shout when the build finishes
11:56
<
GitHub64 >
conda-recipes/master 3701dca whitequark: pyqt5: use prefixed qmake.
11:56
<
sb0_ >
another solution is to remove the pyqt dependency from quamash...
11:57
<
whitequark >
that's stupid
11:57
<
sb0_ >
whitequark, I see no trace of quamash in anaconda, except ours
11:57
<
whitequark >
what does it fucking mean "qt5-qmake is not an executable"
11:59
<
whitequark >
bb-m-labs: force build --props=package=pyqt5 conda-lin64
11:59
<
bb-m-labs >
build forced [ETA 20m13s]
11:59
<
bb-m-labs >
I'll give a shout when the build finishes
11:59
<
GitHub166 >
conda-recipes/master a94086f whitequark: pyqt5: use prefixed qmake.
12:00
<
whitequark >
hm, why did the Windows build fail now...
12:00
<
whitequark >
qbasicfontdatabase.obj : error LNK2019: unresolved external symbol FT_New_Face referenced in function "public: static class QStringList __cdecl QBasicFontDatabase::addTTFile(class QByteArray const &,class QByteArray const &)" (?addTTFile@QBasicFontDatabase@@SA?AVQStringList@@AEBVQByteArray@@0@Z)
12:01
<
whitequark >
curious
12:04
<
whitequark >
bb-m-labs: force build --props=package=pyqt5 conda-lin64
12:04
<
bb-m-labs >
build forced [ETA 20m13s]
12:04
<
bb-m-labs >
I'll give a shout when the build finishes
12:04
<
whitequark >
what the fuck
12:07
<
whitequark >
bb-m-labs: force build --props=package=pyqt5 conda-win64
12:07
<
bb-m-labs >
build forced [ETA 26m30s]
12:07
<
bb-m-labs >
I'll give a shout when the build finishes
12:07
<
GitHub185 >
conda-recipes/master cb0fa81 whitequark: qt5: add.
12:07
<
GitHub185 >
conda-recipes/master b565193 whitequark: pyqt: rename to pyqt5 again.
12:07
<
GitHub185 >
conda-recipes/master ff91807 whitequark: pyqt5: use prefixed qmake.
12:39
<
whitequark >
bb-m-labs: force build --props=package=qt5 conda-win64
12:39
<
bb-m-labs >
build forced [ETA 26m30s]
12:39
<
bb-m-labs >
I'll give a shout when the build finishes
13:31
<
GitHub70 >
misoc/master 0e44624 Florent Kermarrec: tools/flterm.py: first cleanup pass
13:38
evilspirit has quit [Ping timeout: 244 seconds]
13:39
evilspirit has joined #m-labs
13:46
evilspirit has quit [Ping timeout: 250 seconds]
13:47
ylamarre has joined #m-labs
13:48
<
GitHub174 >
artiq/master 28c4d8c Sebastien Bourdeauducq: applets: attempt at fixing embedding bugs on Windows
13:49
<
whitequark >
failed 33 minutes into the build. this is better!
13:49
<
whitequark >
c:\slave64\miniconda\conda-bld\work\qt-everywhere-opensource-src-5.5.1\qtimageformats\src\3rdparty\libwebp\src\dsp\upsampling_sse2.c : fatal error C1083: Cannot open compiler generated file: '': Invalid argument
13:50
<
whitequark >
... did it run out of disk space or what
13:50
<
whitequark >
i couldn't possibly have
13:50
<
whitequark >
everything is awful
13:51
<
whitequark >
I've bumped into the filename length limit
13:52
<
whitequark >
wait, no, I've not
13:52
<
whitequark >
it's trying to open the file with an empty name
13:53
<
whitequark >
but... why
13:57
<
whitequark >
... let's assume it's transient
13:57
<
whitequark >
bb-m-labs: force build --props=package=qt5 conda-win64
13:57
<
bb-m-labs >
build forced [ETA 26m30s]
13:57
<
bb-m-labs >
I'll give a shout when the build finishes
14:25
ylamarre has quit [Ping timeout: 250 seconds]
14:30
<
whitequark >
nope, always fails at the same place
14:37
ylamarre has joined #m-labs
15:06
evilspirit has joined #m-labs
15:36
<
sb0_ >
bb-m-labs, force build artiq
15:36
<
bb-m-labs >
build forced [ETA 3m42s]
15:36
<
bb-m-labs >
I'll give a shout when the build finishes
15:43
<
sb0_ >
okay I'm in for another round of wordpad patching...
15:43
<
whitequark >
what for?
15:44
<
sb0_ >
trying out my applet patch
15:44
<
whitequark >
just install sublime already
15:44
<
whitequark >
actually I think I already did
15:58
ylamarre has quit [Ping timeout: 240 seconds]
15:59
<
whitequark >
lol savannah is still down
15:59
<
whitequark >
morons
16:07
<
sb0_ >
oooh, and another asyncio bug, it seems
16:11
ylamarre has joined #m-labs
16:13
<
sb0_ >
mh. not exactly a bug, just some weird corner case
16:23
<
sb0_ >
no, it's a bug. you cannot shutdown a named pipe server without a race condition
16:23
<
sb0_ >
let me see if I can produce a program that causes the bug most of the time, so I can clearly complain
16:27
evilspirit has quit [Ping timeout: 240 seconds]
16:39
<
GitHub163 >
artiq/master 5f5427f Sebastien Bourdeauducq: protocols/pipe_ipc: work around race condition with loop.start_serving_pipe/close on Windows
16:48
ylamarre has quit [Quit: ylamarre]
16:50
<
GitHub32 >
misoc/master ddac6ed Florent Kermarrec: tools/flterm.py: another cleanup pass
16:52
<
GitHub139 >
misoc/master 653a990 Florent Kermarrec: tools/flterm.py: another cleanup pass
16:54
rjo has quit [Read error: Connection reset by peer]
17:13
<
GitHub4 >
artiq/master 3b81dd5 Sebastien Bourdeauducq: gui: workaround for Qt failing to embed applets in hidden and detached QDockWidgets
17:45
<
cr1901_modern >
sb0_: If you play with my current commit, you'll notice that Fit To View and Zoom To Fit aren't idempotent when toggled in succession. I know why this is/am thinking about the best way to solve it. Short version is that it has to do with the slider widths screwing up the return values from Qt-provided functions.
17:51
<
sb0_ >
bb-m-labs: force build --props=package=pyqtgraph conda-lin64
17:51
<
bb-m-labs >
build forced [ETA 20m13s]
17:51
<
bb-m-labs >
I'll give a shout when the build finishes
18:01
<
sb0_ >
so, after all this bughunt, we can run and plot flopping_f_simulation on windows ...
18:02
<
sb0_ >
let me try talking to the core device
18:05
<
sb0_ >
what, it works?
18:20
kristianpaul has quit [Quit: Reconnecting]
18:21
kristianpaul has joined #m-labs
18:21
kristianpaul has joined #m-labs
18:22
rohitksingh has quit [Ping timeout: 240 seconds]
18:25
folkert has quit [*.net *.split]
18:27
zoobab has quit [*.net *.split]
18:27
siruf has quit [*.net *.split]
18:27
jaeckel has quit [*.net *.split]
18:27
folkert has joined #m-labs
18:27
early has quit [*.net *.split]
18:27
cr1901_modern has quit [*.net *.split]
18:27
zoobab has joined #m-labs
18:27
siruf has joined #m-labs
18:27
early has joined #m-labs
18:28
cr1901_modern has joined #m-labs
18:28
jaeckel has joined #m-labs
18:29
rjo has joined #m-labs
18:29
jaeckel has quit [Changing host]
18:29
jaeckel has joined #m-labs
18:36
rohitksingh has joined #m-labs
18:40
rohitksingh has quit [Ping timeout: 244 seconds]
18:44
rohitksingh has joined #m-labs
19:30
rohitksingh has quit [Quit: Leaving.]
20:08
zoobab has quit [Ping timeout: 276 seconds]
20:10
zoobab has joined #m-labs
20:55
acathla has quit [Read error: Connection reset by peer]
20:57
acathla has joined #m-labs
20:57
acathla has quit [Changing host]
20:57
acathla has joined #m-labs
21:04
zoobab has quit [Ping timeout: 276 seconds]
21:08
zoobab has joined #m-labs
21:56
zoobab has quit [Ping timeout: 276 seconds]
22:04
zoobab has joined #m-labs
22:53
zoobab has quit [Ping timeout: 248 seconds]
23:12
zoobab has joined #m-labs