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
aeris has quit [Ping timeout: 250 seconds]
aeris has joined #m-labs
sh[4]rm4 has joined #m-labs
sh4rm4 has quit [Ping timeout: 265 seconds]
fengling has joined #m-labs
aeris has quit [Ping timeout: 245 seconds]
kristianpaul has quit [Remote host closed the connection]
aeris_ has quit [Quit: en a pas]
kristianpaul has joined #m-labs
aeris has joined #m-labs
Bertl_zZ is now known as Bertl
aeris_ has joined #m-labs
nicksydney has joined #m-labs
<GitHub127> [migen] sbourdeauducq pushed 1 new commit to master: http://git.io/p4V3
<GitHub127> migen/master 555c444 Sebastien Bourdeauducq: mibuild/sim/dut_tb: fix permissions
fengling has quit [Ping timeout: 252 seconds]
<sb0> at the moment ssl won't work at all with proactoreventloop (which is needed to support subprocesses)
nicksydney has quit [Remote host closed the connection]
nicksydney has joined #m-labs
antgreen has quit [Ping timeout: 255 seconds]
ccube has joined #m-labs
<GitHub52> [migen] enjoy-digital pushed 1 new commit to master: http://git.io/pRCD
<GitHub52> migen/master 00e8616 Florent Kermarrec: mibuild/sim: clean up (thanks sb)
<rjo> sb0: ack. i would have thought that ssl can be handled above all socket/asyncio/transport layers sacrificing some integration with the os.
<sb0> yeah, meh
<sb0> rjo, btw regarding controller termination. since signals are implemented in a stupid way in python, and their compatibility with windows is dubious, I propose 1) attempting to terminate via a regular RPC request 2) if that does not work e.g. times out, use subprocess.kill() that sends SIGKILL on Linux and calls TerminateProcess() on Windows
<sb0> ysionneau, how is influxfb?
<sb0> *influxdb
<ysionneau> seems to work nicely, I was able to do a draft of program which subscribes to parameters and then send time+param+value to the influxdb
<ysionneau> now I'm thinking about how to do this reliably
<ysionneau> like you said to reconnect when connection is lost
<ysionneau> also, I'm wondering if using influxdb can block the asyncio loop or not
<sb0> i guess it doesn't matter if some parameters updates are lost when at least one of the connections is defective
<sb0> i.e. transfer from pdb to influxdb if and only if both the pdb and influxdb connections are active, and maintain them in the background
<ysionneau> alright
<sb0> it can, but a quick web search turned up an asyncio influxdb client iirc
<ysionneau> oh!
<ysionneau> I didn't find that
<ysionneau> about the GUI, why thinking about migrating to Qt?
<sb0> ah, no, sorry that was for another database...
<ysionneau> redis has an asyncio client it seems
<sb0> rjo thinks plotting libs are better for qt. Ben from Oxford claims making a workspace environment is easy with qt. I'm waiting to see an implementation of #2....
<sb0> also, there's asyncio support that supports windows and looks well coded
<sb0> and qt is installed by conda by default
<sb0> yeah lots of databases have asyncio clients. but not influxdb it seems...
<sb0> is the influxdb protocol complicated?
<ysionneau> that's just a web API
<ysionneau> with POST requests
<sb0> ah. then you can rebuild it on top of aiohttp.
<ysionneau> you can even just play with curl
<ysionneau> ok
<sb0> just pushing data into a db should be straightforward
<ysionneau> ok I'll have a look
<ysionneau> About "workspace", to be sure I understand well, are you talking about this kind of widget: http://qt-project.org/doc/qt-4.8/qmainwindow.html#qt-main-window-framework ?
<sb0> regarding master connection persistence with asyncio: the controller manager implements that
<ysionneau> ok!
<sb0> yeah, dock widgets, dragging and dropping them, organizing them into different windows, making tabs in windows, managing splits within windows
<ysionneau> I see, to allow user to manage the widgets like he/she wants
<sb0> the multi-window approach is necessary because typical setups run on multiple computer screens
<ysionneau> like modern IDE
aeris_ has quit [Ping timeout: 244 seconds]
<sb0> Ben said he knew how to do that, let's see what he comes up to some weeks from now ... :-)
<sb0> *with
<sb0> if that works well, it would be a deciding factor in favor of qt imo
<ysionneau> I see
<sb0> Gimp also does something like that, so it could be possible with gtk as well, but I don't know how painful that is
<ysionneau> ah yes there is some asyncio event-loop for Qt https://github.com/harvimt/quamash nice!
<sb0> yeah
<sb0> and it's not messy like gbulb
<ysionneau> if it already supports windows: bingo
<sb0> it does
<ysionneau> 17:52 <@mclasen> fallen: not a 1-1 equivalent, no. but you can easily put the pieces together yourself <= ok indeed it seems a bit more painful to do the same in Gtk
<ysionneau> (workspace stuff)
<sb0> there's this gdl library https://wiki.gnome.org/Projects/Vala/GdlSample
<sb0> this gdl thing is in ubuntu
* ysionneau just built the vala/gdl example on mac, works fine
<ysionneau> a bit weird to manipulate though
<sb0> does it work with python
<sb0> ?
<ysionneau> no idea, I just ran the Vala example with valac
<ysionneau> ah, there is python-gdl in ubuntu
sh[4]rm4 has quit [Remote host closed the connection]
<ysionneau> and in macports but only for py26/pè27
<sb0> what's the point of this gobject introspection thing if it doesn't automatically build python bindings?
<ysionneau> ah python-gdl was on old ubuntu
<ysionneau> I guess it should work with pygobject
sh4rm4 has joined #m-labs
sh4rm4 has quit [Remote host closed the connection]
sh4rm4 has joined #m-labs
<sb0> libgdl seems to have small display glitches on my hi-res screen
nicksydney has quit [Quit: No Ping reply in 180 seconds.]
nicksydney has joined #m-labs
<sb0> otherwise it may be fine, if it works with python and windows
<ysionneau> yes but then there is still the gbulb issue
<rjo> ysionneau: use the udp transport when writing to influxdb that is non-blocking.
<sb0> er, you need some XML description of the binding called "GIR"
<sb0> and all the available ones are in /usr/lib/girepository-1.0
<sb0> there's a package gir1.2-gdl-3 that provides it for gdl
<sb0> so it's usable, at least with linux
<rjo> ysionneau, sb0: ignore the influxdb-python package. that's horrible. a minimal client looks something like : http://paste.debian.net/160606/
<rjo> that is easy in asyncio and you only need the udp part.
<sb0> ysionneau, the lda hid cleanup code is still missing from artiq git head.
<sb0> ditto tcube with serial close
<ysionneau> rjo: thanks!
<ysionneau> sb0: by cleanup you mean the hid_close?
<sb0> yes
lrak has quit [Ping timeout: 264 seconds]
<sb0> having a server listening to two IP addresses (and no other ones) with asyncio is a pita.
<sb0> in fact it's impossible unless you write your own event loop
<sb0> blergh
mumptai has quit [Ping timeout: 256 seconds]
mumptai has joined #m-labs
<GitHub198> [artiq] sbourdeauducq created patched_python (+1 new commit): http://git.io/pE66
<GitHub198> artiq/patched_python d19f75c Sebastien Bourdeauducq: protocols/asyncio_server: always bind to localhost by default...
travis-ci has joined #m-labs
<travis-ci> m-labs/artiq#47 (patched_python - d19f75c : Sebastien Bourdeauducq): The build failed.
travis-ci has left #m-labs [#m-labs]
methril has quit [Ping timeout: 252 seconds]
methril has joined #m-labs
methril has quit [Ping timeout: 244 seconds]
methril has joined #m-labs
balrog has quit [Ping timeout: 252 seconds]
balrog has joined #m-labs
methril has quit [Quit: Leaving]