Use get_ext_fullpath instead of get_outputs in ffiplatform._build - #273
Open
afonsojanu wants to merge 2 commits into
Open
Use get_ext_fullpath instead of get_outputs in ffiplatform._build#273afonsojanu wants to merge 2 commits into
afonsojanu wants to merge 2 commits into
Conversation
get_outputs() returns every output the build_ext command produced, not just the one we asked it to build. When the Distribution/build_ext state ends up carrying more than one extension (as reported in python-cffi#246, where the caller's own project already has other native extensions registered), the unconditional [soname] = cmd_obj.get_outputs() unpacking crashes with a bare ValueError instead of naming the extension we actually wanted. get_ext_fullpath(ext.name) asks build_ext directly for the path of the specific extension we passed in, so it no longer depends on exactly one output existing. This also gives a clearer failure mode for python-cffi#229, where a beginner hit the same unpacking crash with no way to tell what went wrong.
mattip
reviewed
Sep 1, 2026
mattip
left a comment
mattip
left a comment
Contributor
There was a problem hiding this comment.
One small nit. LGTM, if CI passes.
| # force a fresh module name/compile every run, so the monkeypatched | ||
| # get_outputs() above is actually exercised instead of reusing a | ||
| # previously-built module cached under the same checksum-derived name | ||
| tag = binascii.hexlify(os.urandom(8)).decode() |
Contributor
There was a problem hiding this comment.
This seems simpler and does not require importing binascii.
Suggested change
| tag = binascii.hexlify(os.urandom(8)).decode() | |
| tag = os.urandom(8).hex() |
Simpler, and drops the now-unneeded binascii import.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closes #246. Related to #229, which hit the same underlying crash.
_build()compiles a single extension and then unpacks its output with[soname] = cmd_obj.get_outputs(). That works as long asbuild_extreports exactly one output, butget_outputs()reflects everything the command built, not just the extension we asked for. #246 shows a project with another native extension already registered ending up with two entries there, so the unpacking raises a bareValueError: too many values to unpackinstead of anything actionable. #229 is a beginner hitting the identical crash with no clue what went wrong, which is basically the same request: give a real answer instead of an opaque unpack failure.build_ext.get_ext_fullpath(name)is the standard way to ask for the output path of one specific extension, so swapping to that sidesteps the whole problem: we get back exactly the file for the extension we passed in, regardless of what elseget_outputs()would have reported.Added a regression test in
testing/cffi0/test_platform.pythat monkeypatchesbuild_ext.get_outputsto return an extra, nonexistent path (simulating the multi-extension situation from #246) and checks thatffi.verify()still compiles and loads correctly. Confirmed it reproduces the exact reportedValueError: too many values to unpack (expected 1)against the unpatched_build(), and passes with the fix. Ran the fulltesting/cffi0/test_platform.py,test_zdistutils.py,test_verify.pyandtest_verify2.pysuites locally, all green.