Control the use of the cl4py readtable. #9
Conversation
|
This is a great suggestion. I will try to merge, test and upload the fix this week. |
|
Probably a test would be a good idea -- is |
|
Sure, any tests would be much appreciated. With some embarrassment I have to admit that, so far, my tests consisted of me running the examples in the README by hand before uploading a new version to PyPi. |
|
Unfortunately, I really don't know how Travis works or how to set up tests that would involve SBCL. But I could see about pytest, and maybe you or someone else could make it work with Travis. |
Verify that the previous version of cl4py would experience errors because the python readtable was used to read *all* input to the lisp process (instead of only input from the Python process).
|
OK, now there are tests! |
It was leaking state into the compiler, and breaking lots of code.
This is needed to be able to write the tests.
New pytests. Added before adding tests for this MR.
Verify that the previous version of cl4py would experience errors because the python readtable was used to read *all* input to the lisp process (instead of only input from the Python process).
rpgoldman
force-pushed
the
fix-readtable
branch
from
da478a9 to
a887be4
Compare
|
Merged. Thank you for all those contributions! |
Glad I could help! Thank you for the library! |
The cl4py readtable is wrapped around everything done by Lisp running inside cl4py, instead of just being wrapped around reading from the stream of inputs from python.
In particular, this means that any lisp code compiled by cl4py is read with the cl4py readtable, which can cause breakage if the standard readtable is expected.
This tweak makes the cl4py readtable apply only around the
readrather than thereadand theeval, as before.