02:12 <GitHub130> [artiq] sbourdeauducq pushed 1 new commit to master: http://git.io/bIa2

02:12 <GitHub130> artiq/master 3a784e3 Sebastien Bourdeauducq: doc/writing_a_driver: use logging.getLogger

03:39 <GitHub4> [artiq] jboulder opened pull request #9: a) lda class is module, set __init__.py properly b) cleanup pdq2's __init__.py (master...example_device) http://git.io/bIxk

04:02 <rjo> couple hours of git tutorial for joe...

04:03 <rjo> sb0: there is a little patch in https://github.com/nist-ionstorage/artiq/commit/021a878d3601ed6a161a290abbb99286c82093d7

04:04 <rjo> sb0: what do you think about the ci stuff?

04:05 <rjo> sb0: if you merge it i'd be happy to take care of it

04:05 <rjo> i might even get around to rigging something with hardware in the loop,

04:06 <rjo> and it does unittest coerave checking now: 38%

04:06 <rjo> coverage...

04:28 <GitHub83> [artiq] sbourdeauducq pushed 1 new commit to master: http://git.io/bLkA

04:28 <GitHub83> artiq/master 78fc5a8 Robert Jordens: setup.py: it's python-datetutil in pypi etc

04:29 <sb0> rjo, I'll have a look. debugging zero_freq_modes with Joe atm

04:35 <sb0> rjo, speaking of HITL. should we have something that measures the dds/pdq/... outputs?

05:42 <rjo_> sb0: finding a capable input device for dds and pdq is difficult/inefficient.

05:42 <rjo_> pdq2 is simulated better, and i would only loop back its digital channel

05:44 <rjo_> and dds readback should be good enough. the rest requires complicated setups. but if there are fast adcs in the future, this would be natural.

05:45 <rjo_> i am pretty much done with refactoring/making the pdq2 host driver nicer and smoother. see nist-ionstorage/pdq2.

05:46 <rjo_> ah. and the other cool thing one can do with travis-ci/xilinx is deploying and publishing the built bitstreams through github releases. that gets xilinx out of the way for quickly getting artiq running.

05:49 <sb0> ah, nice. I've been confused about the pdq2...

05:53 <sb0> Joe finally got his experiment to run in the GUI with realtime plotting, I guess the next step is HDF5 file creation

05:55 <rjo_> and another comment (i have said this before): last time i looked at gtk for windows is was a hopeless mess. maybe that has gotten much better now. but i suspect that qt might still be much easier as a complete package. especially given that there are good and rich plotting libraries for it.

05:58 <sb0> I had a look at those plotting libraries, and they're big and complex packages

05:59 <sb0> cairoplot is small and fast, and well integrated with python

06:03 <sb0> also qt uses c++, which is worse than gobject

06:03 <rjo_> pyqtgraph was a breeze for me.

06:03 <rjo_> you don't see anything of c++.

06:04 <rjo_> and i would bet that pyqtgraph beats the crap out of cairoplot in speed.

06:05 <rjo_> but all this is irrelevant if the gtk/glib/gir stack does not fly on windows.

06:07 <sb0> why not? I remember installing gimp on windows, and it went fine

06:08 <sb0> that was a while ago though...

06:10 <rjo_> i have not tried gimp for a while. just a gtk/glib/pygtk stack was a major pain/impossible.

06:10 <sb0> what went wrong?

06:11 <rjo_> conda does not have it afaict.

06:11 <rjo_> oh. i could not find a combination that was new enough and workd together with itself.

06:12 <sb0> http://sourceforge.net/projects/pygobjectwin32/files/?source=navbar

06:12 <rjo_> gtk2 vs 3, glib versions, vs compiler mingw32 versions. could not compile it under visualstudio at all. pygtk did not like it.

06:15 <rjo_> maybe it is all better now. but we need to make sure before starting to fine tune widget positions in gtk. also need to ensure cairoplot is rich enough. seems rather slim compared to what i usually do for plotting but we need to be thorough.

06:56 <sb0> if all else fails, on windows we can package a mingw gui with its own python based on this sourceforge project, and the rest using conda/visual studio

14:34 <GitHub84> [artiq] sbourdeauducq pushed 1 new commit to master: http://git.io/bme5

14:34 <GitHub84> artiq/master fc92a66 Yann Sionneau: asyncio: Use ProactorEventLoop on Windows

19:13 <sturmflut> sb0: I did a complete reinstall of the Thinkpad Tablet 8 with Ubuntu 14.10 (kernel 3.16.0-23-generic). eMMC works properly, so the propblem must have been introduced with the more recent kernels or Ubuntu has patched/configured something.