sb0 changed the topic of #m-labs to: ARTIQ, Migen, MiSoC, Mixxeo & other M-Labs projects :: fka #milkymist :: Logs http://irclog.whitequark.org/m-labs
fengling has joined #m-labs
attie has quit [Ping timeout: 272 seconds]
attie has joined #m-labs
fengling has quit [Quit: WeeChat 1.2]
<GitHub193>
[artiq] sbourdeauducq pushed 1 new commit to master: http://git.io/vWGR8
<GitHub193>
artiq/master e4165ae Sebastien Bourdeauducq: gui/displays: do not test for empty list using bool conversion (breaks for numpy arrays). Closes #153
travis-ci has joined #m-labs
<travis-ci>
m-labs/artiq#529 (master - e4165ae : Sebastien Bourdeauducq): The build has errored.
<whitequark>
if you look below, it actually uses ctypes to patch cpython's internal data
<sb0>
ok...
<whitequark>
I think I'll stick with the current approach, where I just put it into exception message
<sb0>
right now the main problem with your compiler is that it doesn't run the current experiments, so I'd move implementing this hack down the bottom of the priority list
<sb0>
ok
<GitHub49>
[migen] whitequark pushed 3 new commits to new: http://git.io/vWZwQ
<GitHub49>
migen/new 1890b0c whitequark: conda: also add build number, not just string.
<GitHub49>
migen/new 380018b whitequark: conda: put git hash back build string.
<GitHub49>
migen/new 44e5c68 whitequark: conda: restrict python to 3.5.* explicitly.
travis-ci has joined #m-labs
<travis-ci>
m-labs/migen#153 (new - 44e5c68 : whitequark): The build passed.
<whitequark>
sb0: is there a reason you do not run tests within conda, but do conda install and then python setup.py test instead?
<sb0>
I didn't write that code, and I don't know
<sb0>
I think rjo did
<whitequark>
ok
<whitequark>
oh, that's because it doesn't work; setup.py is not installed
<cr1901_modern>
sb0: On the issue https://github.com/m-labs/artiq/issues/128, rjo mentions wanting support for a *fancier* scanning widget. Was there a simpler widget in the past?
<sb0>
cr1901_modern, well, the spinboxes.
<sb0>
why does a templating engine need to fake exceptions?
<cr1901_modern>
Hmmm, the only spinboxes I've seen so far are due date and priority.
<sb0>
cr1901_modern, try ArgumentsDemo
<cr1901_modern>
Yea, I just figured that out. Turns out all the demos are bombing with errors (Note to self: don't redirect output to a file if you're not gonna read it)
<sb0>
what errors?
<sb0>
cr1901_modern, ^
<cr1901_modern>
artiq.master.worker.WorkerError: Worker ended while attempting to receive data
<cr1901_modern>
(sorry, was afk)
<sb0>
that's not all
<cr1901_modern>
and still am
<cr1901_modern>
I'll paste the log
<GitHub111>
[artiq] whitequark pushed 3 new commits to master: http://git.io/vWZyv
<cr1901_modern>
whitequark: Have you had luck building h5py with Visual Studio 2015 on Windows
<cr1901_modern>
>*
<cr1901_modern>
?**
<whitequark>
I have never tried
<cr1901_modern>
Here's the problem: Python 3.5 requires Visual Studio 2015 (or VC14 as it's sometimes called). All external python packages also need to be built with VC14. I got PyQt/Qt working with VC14, so I can upload them to conda at any time. But h5py has been refusing to build, and the logs not all that helpful.
<cr1901_modern>
It doesn't tell me what errors it found- just that there were errors.
<cr1901_modern>
The remaining part of the log is all test output
<cr1901_modern>
(of course half of them failing b/c ctest apparently isn't smart enough to not run if the build fails)
<cr1901_modern>
Ah well, let me try harder lol
<cr1901_modern>
Okay looks like ctest has a -VV command switch that I didn't use (the log was generated using -V), so hopefully this gives me some hints