PyX.jl/src, branch pyrecwrap[no description]https://git.bnewbold.net/PyX.jl/atom?h=pyrecwrap2016-04-05T15:29:56+00:00BROKEN: start integrating pyrecwrap2016-04-05T15:29:56+00:00bnewboldbnewbold@robocracy.org2016-04-05T15:29:54+00:00urn:sha1:819a184c816342abc258ec0f905707ed4b7b44af
Several test (eg, all `style.linewidth(0.2)` calls) fail.
The problem is that python classes can be called like a function
("style.linewidth(0.2)") or accessed like a module
("style.linewidth.THICK"), while Julia modules/functions can not.
Revert "pyrecwrap: skip some wrapping as an optimization"2016-04-04T22:33:04+00:00bnewboldbnewbold@robocracy.org2016-04-04T22:33:04+00:00urn:sha1:37d2c0d3749461ef86b0886f27128a88ddcc2f86
Wasn't worth the complexity; hard to get unittest to work because
results (eg, for "path") were being cached.
pyrecwrap: skip some wrapping as an optimization2016-04-04T22:32:51+00:00bnewboldbnewbold@robocracy.org2016-04-04T22:22:09+00:00urn:sha1:3a3feee057933c691e7184ef82c62ec705d4ce78
This cuts down on the number of wrapped Julia Modules by ~30%. Might not
be worth it.
pyrecwrap: nest python classes as modules2016-04-04T22:07:21+00:00bnewboldbnewbold@robocracy.org2016-04-04T22:05:17+00:00urn:sha1:9b2d9f990537b8c36c0e3fd936b4ae14b88a59ab
This brings the behavior closer to how PyCall's pywrap() itself works,
and makes this function useful for PyX.
With Python3 and PyX 0.14.1, this generates 1100 nested modules or
classes; the previous commit (modules only) had much fewer (167
modules).
add recursive version of PyCall's pywrap2016-04-04T21:32:24+00:00bnewboldbnewbold@robocracy.org2016-04-04T21:32:24+00:00urn:sha1:6ca0769a2f82c7c7d6a34fda2571b9ae15950842
Will be used to have deeply nested Julia Modules for deeply nested
Python modules Eg, os.path.genericpath.* in python becomes
os.path.genericpath.* in Julia.
This version only works with modules, not Python classes.
Thanks to James Porter at the Recurse Center for helping with this!
change pipeGS's 'device' from kwarg to arg2016-03-03T20:26:20+00:00bnewboldbnewbold@robocracy.org2016-03-03T20:16:58+00:00urn:sha1:d5d207f0a13b883a408fa6a7a948cabf1754c9b3fix png and jpeg MIME; export writePSfile and writeGSfile2016-03-02T05:57:32+00:00bnewboldbnewbold@robocracy.org2016-03-02T05:57:28+00:00urn:sha1:d6c2ac682165bfc4c7ecf9fb7027206735e5bfef
writePSfile accidentally wasn't wrapped.
Unlike the other file writing functions, writeGSfile requires a file
path, can't pass an IO object or file descriptor. Enforce this with a
type annotation (AbstractString).
The above means that png and jpeg writemime() support requires using
pipeGS instead of writeGSfile.
huge rework adding MIME output support, PyxCanvas, wrapper, more2016-02-23T21:14:17+00:00bnewboldbnewbold@robocracy.org2016-02-23T21:14:13+00:00urn:sha1:fa8f048e78bb9965902371f937d761267bd6b63b
Apes PyPlot to map a new Julia type (PyxCanvas) to the PyX python
"canvas" class, and defines many methods on this new type, including
MIME output.
Also addes ghostscript piping stuff, and corrects Python3 (SVG) support.
This now at least sort-of works with Jupyter.
Probably needs a refactor/cleanup, definately needs mime-specific tests,
and the Graph object should also be wrapped (PyxGraph).
add conditional (py3) writeSVGfile function export2016-02-22T02:45:17+00:00bnewboldbnewbold@robocracy.org2016-02-22T02:45:17+00:00urn:sha1:ba0bb2fc756deaa8f02cc58148fc4ef9095e85ffyet more exports; change style_line* exports to be consistent2016-02-22T02:07:35+00:00bnewboldbnewbold@robocracy.org2016-02-22T02:07:35+00:00urn:sha1:5dab103e3ed4728b6e920eb88b9cedc86dfc3183