Fix fsevents segfault #763
Conversation
src/watchdog_fsevents.c
Outdated
| const char *c_string_ptr; | ||
| PyObject *py_string; | ||
|
|
||
| c_string_ptr = CFStringGetCStringPtr(cf_string_ref, 0); | ||
| c_string_ptr = CFStringGetCStringPtr(cf_string_ref, kCFStringEncodingUTF8); |
There was a problem hiding this comment.
You can declare and initialize c_string_ptr in the same line.
samschott
force-pushed
the
fix-fsevents-segfault
branch
from
930bf3a to
c993fa4
Compare
this prevents unicode normalisation on HFS+
|
The tests on macOS were all failing but only on Travis. The new test was creating a directory named "TéstClass" with the e-acute represented by U+00E9. On Travis, the resulting native event had a path with name "TéstClass" where the e-acute is U+0065 and U+0301. It appears that the path as reported by FSEvents is correct, the actual file which is created is using the second, decomposed representation of e-acute, contrary to what is passed to A good writeup of the changes: https://eclecticlight.co/2017/04/06/apfs-is-currently-unusable-with-most-non-english-languages/ I have therefore changed the file name to use the decomposed characters in the first place. This explains the last commit with changes that might otherwise appear to be just "cosmetic". I can confirm that the new file name still tests the same code path and that the test fails on the master branch. |
|
Thank you for the fix and for the explanation about unicode stuff :) No worries about your "C level", you are making the module so much better that it does not count :) |
|
Release v2.0.1 online. |
|
Nice! |
Fixes #762 and adds a test case.
Edit: Apologies for the oversight. I did warn you that this was my first time programming in C...