-
-
Notifications
You must be signed in to change notification settings - Fork 30.9k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
LGHT0091: Duplicate symbol 'File:include_pyconfig.h' found #79480
Comments
Build of Python 3.6.7 fails with error Steps to reproduce this error:
NOTE1: if step 5 is skipped, the overall build does not fail, but the PGInstrument x64 Interpreter fails in distutils with error "E:\RepoGiT\3.6\include\Python.h(8): fatal error C1083: Cannot open include file: 'pyconfig.h': No such file or directory", as shown in message https://bugs.python.org/msg329247 of issue https://bugs.python.org/issue35157 NOTE2: if pyconfig.h in step 5 is copied but the environment variable is not created, the PGInstrument x64 Interpreter fails in distutils with error "LINK : fatal error LNK1104: cannot open file 'python36.lib'", and the overall build fails with the same error shown in the title of this Issue. NOTE3: If all the steps are run, the PGInstrument x64 Interpreter succeeds in its tests but the overall build fails with the error shown in the title of this issue. All the verbose of this compilation is included in the file attached. |
I'm going to declare that this is a bug in distutils not correctly setting INCLUDE and LIB values when building in a source directory. When this is fixed, there will be no need to copy pyconfig.h anywhere, and the installer build will be fine. (And it should be noted that we don't rely on 100% pass rate of the test suite when training for PGO - the point is to exercise as many code paths as possible. The test suite has shown to be as good as any other generic workload, and is easily available in our repo.) |
I want to document a workaround if anyone benefits from it, this would be step 7:
This is only a workaround and requires the user placing attention to the build process (maybe for the first 5-10 minutes). Having the option "-q" removed from the buildrelease.bat file (as discussed in issue https://bugs.python.org/issue35739 ) helps with determining when to safely remove the file so that the distutils test pass. |
My PR fixes both issues properly, but adds a minor "feature" to distutils by making distutils.sysconfig.get_python_inc() potentially return a set of include paths (with os.path.pathsep separators). This is fully internal, so right now we know it'll only occur when building in the source tree on Windows, so I'm not concerned about it. It'd also work on any platform through an environment variable, which it is, but it looks like most people simply attach "-I" at the start and pass it to gcc (which again, since they're not on Windows, they're not going to hit the one case we know about). I'd rather do it properly than have a one-time hack just for this build scenario, and I certainly don't want to be copying files within the source tree. So I'm intending to go ahead with this design, but wanted to allow any interested parties to push back if they know of some reason that I don't. |
Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.
Show more details
GitHub fields:
bugs.python.org fields:
The text was updated successfully, but these errors were encountered: