setting using asymptote for graphics by default #25
Conversation
mmatera
changed the title
| @@ -49,7 +49,7 @@ def read(*rnames): | |||
| is_PyPy = platform.python_implementation() == "PyPy" | |||
|
|
|||
| INSTALL_REQUIRES = [] | |||
| DEPENDENCY_LINKS = [] | |||
| DEPENDENCY_LINKS = ["http://github.com/Mathics3/pymathics-asy#egg=pymathics-asy"] | |||
There was a problem hiding this comment.
Please don't do this. We want to reduce complexity and the number of dependencies and this just adds this back in.
The front ends should allow for reading profiles and customization.
And the mathics-ominbus project was created to pull in all of the different pieces.
There was a problem hiding this comment.
OK, then I would discard this PR. However, I think this would be a suitable (temporary) patch to fix the issue with graphics, this is why I put it here.
The other way to solve the graphics issue would be to update and debug all the javascript machinery.
There was a problem hiding this comment.
We should probably think about and discuss a strategy, short-term and long term.
I confess that in a way I did I think the same kind of thing with pymathics-graphics that you are doing here. So I understand and am sympathetic to short-term workarounds.
Personally, I am happier with short-term workarounds only after we have a clear idea of what the longer-term strategy is.
A long while ago, it was my impression that we were going to eventually ditch Django and go with some sort of Jupyter-based solution. At least in my mind, that hasn't been feasible yet because the foundations of the Mathics core and how it interacts with "front-ends" just doesn't feel fully worked out, although we are in my opinion making good progresss toward that end,.
There was a problem hiding this comment.
OK, I agree with that, and I thought that the strategy was more or less this one:
-
Fix graphics in Django with a plugin (pymathics-matplotlib or this one) that works in a very short term, in order to have a functional candidate for release soon. I would also prefer to use Matplotlib instead Asymptote for rendering graphics, but it happens that the development of the asy interface is much more advanced than the mpl one. This is why I have used it in this PR.
-
Use this to improve the way in which graphics packages are loaded and works. Regarding this, working on pymathics-asy I think I have discovered the root of the problem with graphics: it is rooted in the javascript/MathJax postprocessing. To really fix it we should look at the javascript routines that perform the XHTML replaces (mathics.js). We will need to do that if we want to have svg graphics. By now, I solve this by sending png graphics.
-
In the long term, to use what we learned working on that plugin to improve the jupyter's interface, which in the long term replace Django frontend as the main frontend of the project. Yesterday I have been doing some progress in that direction. I am updating this binder: https://mybinder.org/v2/gh/mmatera/nbcoursebase/master with the improvements to make online checks.
There was a problem hiding this comment.
Ok. Thanks for the information and plan!
Let me think about abount this, look at the code, try things, and try to understand the what's been done and the existing code.
|
@rocky , If you can, I would appreciate some help in setting the dependency for the module in setup.py |
Ok I will look at over the weekend. |
rocky
force-pushed
the
master
branch
5 times, most recently
from
f5e0f88 to
96572aa
Compare
rocky
force-pushed
the
master
branch
4 times, most recently
from
0963d16 to
400465e
Compare
This PR adds pymathics.asy module as a dependency and loads it when mathicserver is started. This solves the issue with labels and text inside graphics.