wesm opened a new pull request #7334: URL: https://github.com/apache/arrow/pull/7334
Current manylinux wheel packages on master: * .whl manylinux1 package is **61 MB** * Installed size is **223 MB** This patch * .whl package is **15 MB** * Installed size is **57 MB** That's more than a 4x size reduction. There's several things in this patch: * We no longer ship 2 copies of shared libraries in the wheels. We ship just the SO-versioned shared libraries now. Because this creates problems for linkers (`-larrow -lparquet` etc won't work as is), I added function that tries to create the necessary symlinks when you call `pyarrow.get_library_dirs()`. If pyarrow is installed somewhere where you can't create symlinks and the symlinks don't exist, it will print a message instructing you to run the symlinking function as root. This was the simplest strategy I could think to get out of this mess. * Gandiva is disabled. If we're going to ship Gandiva as a wheel, I think we should do it as an add-on `pyarrow_gandiva` package per ARROW-8518. * Environment variable PYARROW_INSTALL_TESTS added to not install `pyarrow.tests`, which is about 2.3MB uncompressed. I don't think we need to ship the tests in the wheels. * Compiled Cython sources are no longer shipped. I'll need some help kicking the tires on macOS and Windows and to make sure the Crossbow builds are all passing. ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected]
