Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • Jan 31 16:26
    kmuehlbauer commented #1568
  • Jan 31 16:15
    kmuehlbauer commented #1568
  • Jan 31 15:53
    djhoese commented #1568
  • Jan 31 15:46
    kmuehlbauer edited #1568
  • Jan 31 15:44
    kmuehlbauer opened #1568
  • Jan 31 15:32
    djhoese commented #1566
  • Jan 31 15:30
    djhoese commented #1567
  • Jan 31 14:18
    GuillaumeFavelier commented #1561
  • Jan 31 14:15
    GuillaumeFavelier commented #1561
  • Jan 31 12:46
    opekar synchronize #1130
  • Jan 31 09:54
    kmuehlbauer commented #1567
  • Jan 31 09:53
    kmuehlbauer synchronize #1567
  • Jan 31 09:37
    kmuehlbauer labeled #1567
  • Jan 31 09:37
    kmuehlbauer labeled #1567
  • Jan 31 09:37
    kmuehlbauer labeled #1563
  • Jan 31 09:36
    kmuehlbauer labeled #1563
  • Jan 31 09:28
    kmuehlbauer commented #1567
  • Jan 31 09:27
    kmuehlbauer synchronize #1567
  • Jan 31 09:19
    kmuehlbauer assigned #1563
  • Jan 31 09:17
    kmuehlbauer opened #1567
Mark Harfouche
@hmaarrfk
@djhoese the numpy warning is a little annoying. I might try to add a filter to catch it. makes me rethink our warning strategy is scikit-image. as for the point releases, mostly the release process is so drawn out with CIs that people seldom have time to cut releases. It's a little strange that 0.14.4 or .5 will have important patches that aren't available anywhere else. This is mostly because we feel that the manual uploading step is better than auto uploading.
David Hoese
@djhoese

@hmaarrfk I can try to make a vispy release in the next couple days for the numpy warnings. I just got a new laptop so I've been falling behind my work more than usual trying to get everything working the way I want. I want to try updating vispy to use setuptools_scm before releasing though. I'm getting real sick of manually editing the version of the package.

Why do you like manual uploading over auto uploading? I've always been worried about screwing things up by building packages locally (having extra files getting packaged up because the environment/pkg dir isn't clean).

sean.r.breashears-1@ou.edu
@SBreashears_gitlab
Hi, I was wondering if you can use Vispy on integrated graphics such as the Intel HD Graphics 630 that comes alongside an I5? I know it uses OpenGL version 4.5 but am not sure if Vispy requires a dedicated GPU. Thanks!
David Hoese
@djhoese
@SBreashears_gitlab Theoretically it should work fine. We have seen some issues in the past with people trying to switch between dedicated and integrated GPUs on the same system, but I think most of those were PyQt's fault.
Mark Harfouche
@hmaarrfk
@djhoese I use versioneer for my personal (company) projects extensively. It works really well, and enables REALLY fast releases.
The issue with scikit-image is that it predates sensible wheel build systems
so the core team has (rightfully) been weary of them.
I think there is a way to speed up the process, but it would require many core devs to get behind a new system, and I don't think here has been concensus for a while
I also think scikit-image predates things like readthedocs
which made releasing documetnation easy
today, the release model of tag and forget is really put in place
Mark Harfouche
@hmaarrfk
versioneer, readthedocs, and travis keys should be good to have faster release cycles (and this is what dask does I think)
David Hoese
@djhoese
@hmaarrfk fyi versioneer is technically unmaintained. We just merged a PR for vispy today that uses setuptools_scm to accomplish the same thing. We'll see how it goes.
sean.r.breashears-1@ou.edu
@SBreashears_gitlab
Quick question (if this isn't the right place, let me know where to ask questions): Not sure if anyone here uses wxPython, but I'm struggling to embed any kind of plot (or figure) into a wx.Panel or wx.Frame. If anyone here has tried this or might have a quick suggestion let me know, else I'll just keep attempting myself. Thanks!
David Hoese
@djhoese
@SBreashears_gitlab The core vispy developers don't really use Wx, but I've tested the basically funcionality in the past. What version of Wx are you using? Have you tried any of the basic vispy examples to make sure it isn't an OpenGL issue on your system?
sean.r.breashears-1@ou.edu
@SBreashears_gitlab
Version: 4.0.6
Using Vispy version: 0.6.0.dev0
David Hoese
@djhoese
FYI there are some slightly newer versions of vispy. The changes shouldn't really change anything for Wx, but FYI
what about the vispy examples?
@larsoner @kmuehlbauer Any objections to a vispy 0.6.2 release this week?
sean.r.breashears-1@ou.edu
@SBreashears_gitlab
I've been able to get simple_wx.py to run from the github but am not able to reproduce it in an embedded panel. I know I'm missing a connection here somewhere but just haven't wrapped my head around it yet...
good to know there is a newer version though, thanks.
sean.r.breashears-1@ou.edu
@SBreashears_gitlab
A little background: I've been using embedded matplotlib plots inside of multiple wx.Panels to plot real-time incoming insturment data, but have found the lack of speed that matplotlib produces to be a bit too sluggish. I've been able to run many examples on the Vispy main page and have been able to achieve much higher frame rates (even with the poor Intel integrated graphics my laptop has) than matplotlib could ever produce. It just seems that the wx.Panels don't like canvas objects, figure objects or anything else I can throw at it...
Eric Larson
@larsoner
no objections to release for me
Kai Mühlbauer
@kmuehlbauer
@djhoese Nothing to add from my side, so no objections to release.
David Hoese
@djhoese
@SBreashears_gitlab If you can break things down to a simple example that I can run locally I could try to help. Get an example make an issue on the vispy github, however it seems this may be a Wx issue and not necessarily a vispy one.
Mark Harfouche
@hmaarrfk
hmm, interesting that versioneer is unmaintained
David Hoese
@djhoese
@hmaarrfk Yeah, last I checked it hadn't received updates for a while and needed some updates for PEP8 styling or something like that (it was failing our auto-linting so we had to exclude it)
Mark Harfouche
@hmaarrfk
PEP8 is this doubled edged sword. Versioned code like versioneer should simply be excluded.
The myriad of configurations for PEP8 and styling makes me hate using them.
David Hoese
@djhoese
yep, not fun. Best to stick with one that looks good to the team and stick with it
Mark Harfouche
@hmaarrfk
David, when running Vispy from spy
Spyder, it seems to like to default to using the notebook backend.
To change it, you have to use the use function
Which can't be called after Vispy has been invoked.
Meaning your whole analysis has to be restarted.
Anyway to address that?
David Hoese
@djhoese
@hmaarrfk I added logic a couple versions ago to auto-detect when we were in a notebook. Does spyder use a ipython session or a notebook session? Something about the ipython kernel must be making vispy think its in a notebook. We could try to make that more accurate if possible. Otherwise, you can set the app you are using at the top of your module.
I don't think it even needs to be before the other imports
just before the app is started
I have never used spyder so if you could look at and try out the get_ipython function inside their terminal (or whatever environment they are running your vispy code from) and see if there is something that identifies a notebook compared to a spyder terminal that would be great
Mark Harfouche
@hmaarrfk
Yeah. Spyder uses ipython
What would you like to know about the get ipython call
Mark Harfouche
@hmaarrfk
anyway, this is definitely more difficult than a gitter chat: vispy/vispy#1708
ecneufeld
@ecneufeld
Hi out there, i need help. I want to render wafermaps including located structures with vispy embeded in jupyter. Therefore i 'simply' need to create lot's of rectangles at least 100.000 (but in average 30.000 and maximum is up to 5.000.000). Rectangles should be colored. Pan and Zoom as well as click events should be available. So my question is: Should i try to use gl_PointSize in shader for rendering those rectangles or should i try using a lot of visuals.RectangleVisual instances?
Edit: in average: 30.000 different rectangles (size and position). Thats my usecase
ecneufeld
@ecneufeld
vispy (0.6.1) seems to slow when using pan and zoom when trying those examples. I'll go for alternatives such as threejs.
David Hoese
@djhoese
@ecneufeld I would honestly recommend not using vispy for this if jupyter is the target backend. Due to the way vispy is designed it is difficult for us to get good performance out of the jupyter backend as-is and would require large changes. Regardless, the RectangleVisual would not be good for this. Any solution that requires more than 10 visuals will perform terribly as it requires a separate GL program for each visual. Each program is drawn sequentially so it takes the GPU a long time to get through each one. It would be best to have one visual that draws all of these rectangles with the appropriate color (I'd never heard of a wafermap before). From googling I couldn't tell if you would need a 2D pan/zoom surface or a 3D with a "z" elevation to the rectangle positions. Is there a big difference between a wafermap and a regular image? I wonder if the ImageVisual would be a good enough start for you, but like I said the jupyter backend will not perform well and is still experimental in my mind.
Matthew Spellings
@klarh
@ecneufeld not sure how snappy you need it to be, but you could also try plato (using vispy as the backend): https://gist.github.com/klarh/d50a3b81251010f88124517df45beebd