This change moves almost everything in the `upb/` directory up one level, so
that for example `upb/upb/generated_code_support.h` becomes just
`upb/generated_code_support.h`. The only exceptions I made to this were that I
left `upb/cmake` and `upb/BUILD` where they are, mostly because that avoids
conflict with other files and the current locations seem reasonable for now.
The `python/` directory is a little bit of a challenge because we had to merge
the existing directory there with `upb/python/`. I made `upb/python/BUILD` into
the BUILD file for the merged directory, and it effectively loads the contents
of the other BUILD file via `python/build_targets.bzl`, but I plan to clean
this up soon.
PiperOrigin-RevId: 568651768
A couple weeks ago we moved upb into the protobuf Git repo, and this change
continues the merger of the two repos by making them into a single Bazel repo.
This was mostly a matter of deleting upb's WORKSPACE file and fixing up a bunch
of references to reflect the new structure.
Most of the changes are pretty mechanical, but one thing that needed more
invasive changes was the Python script for generating CMakeLists.txt,
make_cmakelists.py. The WORKSPACE file it relied on no longer exists with this
change, so I updated it to hardcode the information it needed from that file.
PiperOrigin-RevId: 564810016
This partially fixes https://github.com/protocolbuffers/protobuf/issues/10088. The test case from that bug significantly improves with this change. However we still have a global map that does not shrink, which can still create the appearance of leaking memory, as it will not be freed until the module is unloaded.
PiperOrigin-RevId: 563124724
This is the second attempt to fix our Git history. This should allow
"git blame" to work correctly in the upb/ directory even though our
automation unexpectedly blew away that directory.
- replace all instances of the deprecated iterator with the much nicer new one
- fix a bug which caused the new iterator to skip all values in the hash array
- fix a bug which caused the new iterator to skip the first value in the hash table
- delete the old iterator code
- also replace most uses of the deprecated string hash table iterator
PiperOrigin-RevId: 489093240
After some debugging I was able to determine that the Win32 DLL was failing to load libgcc_s_sjlj-1.dll. If we statically link libgcc instead, there should be no runtime dependency on libgcc.
PiperOrigin-RevId: 448338965
1. For some reason the version script was not working, it was failing to export the main symbol for the Python extension. I fixed this by using the `visibility` attribute instead to export the `PyInit__message` function.
2. We were not properly stripping the `python/dist/` prefix for the C module, which was making the module exported under the name `dist.google._upb` instead of `google._upb`.
3. The `py_library()` rule was failing to actually propagate the module file into the wheel, so I just removed it.
PiperOrigin-RevId: 445446611
* Fixed some reference leaks.
* Fixed a few reference leaks.
* Fixed a few more memory errors.
* Fixed a few more reference leaks.
* Revert minimal_test.py.
* Re-enable limited API.
* Removed some debugging and spurious changes.
* Addressed PR comments.