Improve Windows coverage and fix Windows tests#1176
Conversation
e182236 to
a988068
Compare
files.py: implement OperatingSystem option
a988068 to
c63e46f
Compare
|
@mmatera there are some fixes in here that might be split ou In Also, the black formatter ( I may try to get actions CI working on this later. |
|
By now, looks very good!
Indeed, today I installed pyflakes, and I found this issue. Thanks for solving it! |
|
Also, maybe we can merge this branch with #1173 with has some similar contributions. |
files.py: implement OperatingSystem option
c63e46f to
3d1815d
Compare
…to windows-compatability
|
|
||
| >> ToFileName[{"dir1", "dir2", "dir3"}] | ||
| = dir1/dir2/dir3 | ||
| = dir1...dir2...dir3 |
There was a problem hiding this comment.
... would be also a good way to deal with digits beyond accuracy in numeric tests!
| evaluation.timeout_queue.pop() | ||
| return failexpr.evaluate(evaluation) | ||
| except: | ||
| if sys.platform != "win32": |
There was a problem hiding this comment.
Is TimeConstrained what is failing? or is the expression we have chosen to test it?
There was a problem hiding this comment.
I believe the implementation is Unix/POSIX specific. I haven't investigated though.
44c3901 to
6c0a8cc
Compare
importexport.py: reinstate some of the changes made before lost due to conflitcs CHANGES.rst: note what we did here lists.py: remove warning due to bad escape character in \ldots
cd9149e to
940ead7
Compare
940ead7 to
0e87781
Compare
|
@mmatera this has been most not fun to work on. I have spent way too much time on this, so I am stopping work on this. Also I am merging this in because coping with the conflicts with master, in particular In general, it appears that doc tests should not rely on unicode output such as were found in Base64Encode and Print. Instead put those under pytest and test for character-set capability. In sum although this is incomplete, it is a start. |
|
@rocky, thanks for all this work! I will think about how to solve the code page issue in the docstrings. |
|
@mmatera It is not about the docstrings per se. It has to do with test output that uses unicode. At some point we will probably have a better way to indicate unicode and/or native character set capabilities. However setting attributes like this is doomed for the docstring system as it is which needs to be replaced by something that is not so custom. However easy it is to add docstring tests, please keep in mind:
Until we use some other less custom system, if a test doesn't fall into this category consider putting it in pytest. |
|
@rocky,completely agree with that criteria. However, I think that it is possible to implement tests using unicode / the character table of the platform. In any case, I think that your work on improving compatibility improves a lot the quality of the code that we have in mathics. |
files.py: implement OperatingSystem option