-
-
Notifications
You must be signed in to change notification settings - Fork 24
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
Attempt to fix the Win32 build. #18
Conversation
OK, well, I'll write up some notes on what seems to be going on here. First, there seems to be a disagreement between the Cygwin settings in the Miniconda distribution, and in the Cygwin tools that are installed during the conda-build process. The Miniconda version of
The
This becomes relevant because the relevant programs later map these Unix paths back into Windows paths. Some Second, the particular build process used for Finally, I'm not sure exactly when it happened, but I think that a somewhat recent version of |
Also, error out earlier while we're at it.
Hi! This is the friendly automated conda-forge-linting service. I just wanted to let you know that I linted all conda-recipes in your PR ( |
For reference, here are relevant the commits in the Git history of
The first commit showed up on May 5, 2016, just before the 1.20.2 release. |
Aha! The trailing semicolon in the definition of I suspect that MSYS2 is parsing the variable into a string of zero length, which then matches against every command-line argument, making it decide to convert no command-line arguments to any Windows programs. I'll file this as a conda-build bug since that's not the intended behavior; the trailing semicolon comes from expanding a possibly-absent previous value of the variable. |
…XCL% The previous method of setting the %MSYS2_ARG_CONV_EXCL% environment variable tried to preserve any preexisting value of the variable. Not only does this make the build process depend on the initial environment more, it generally leads to a trailing semicolon in the variable value. This seems to cause MSYS2 to parse the variable as having an empty value that prevents *all* argument conversions, as investigated in the conda-forge `cairo` package: conda-forge/cairo-feedstock#18. I do not know what fallout this change will have for existing build scripts, but the current behavior cannot be what was intended.
There's some related discussion in this conda-build pull request. |
Great job, this is passing now! |
Could you also remove my suggested hack https://github.com/conda-forge/cairo-feedstock/blob/master/appveyor.yml#L83 as part of this PR? |
Revert 83dba28 and tidy up the fix now that the culprit has been identified.
The CI failures were because I accidentally pushed my branch to the |
Commentary to follow, if this actually works.