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
Fixes to help with conda-forge builds #1591
Conversation
The two values need to be combined as a single string to prevent CMake from interpreting them as a list and inserting a semicolon that results in a broken compiler command that looks like this: g++ -I/some/path;-fopenmp source_file.cpp ...
This test was failing by a slim margin on ppc64le Linux builds
There is another tolerance issue in #1271 (which was resolved by patching downstream) - I'd consider adopting this for the same rationale as provided here for |
I stand by my comment in that PR, and don't think the patch proposed there should be adopted as our baseline. It's not even clear at this point whether that patch is still necessary on those systems, or whether it was just a transient due to specific builds of g++ and/or glibc. |
Sure. I was mainly asking as the tolerance adjustment was pretty minor, going from |
For package builds, do not specify compiler. Otherwise, specify the same compilers that were used to build Cantera. In both cases, allow CC, CXX, etc. environment variables to override the settings in SConstruct.
Codecov Report
@@ Coverage Diff @@
## main #1591 +/- ##
==========================================
- Coverage 70.61% 70.58% -0.03%
==========================================
Files 379 379
Lines 59153 59169 +16
Branches 21252 21257 +5
==========================================
- Hits 41768 41762 -6
- Misses 14311 14331 +20
- Partials 3074 3076 +2
... and 1 file with indirect coverage changes 📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more |
99de898
to
c85fdb5
Compare
Also filter out paths referring to conda build environment
76c5c66
to
2ba2cb0
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks, @speth ... this looks all good to me!
Changes proposed in this pull request
Fixes to resolve issues encountered while trying to update the conda-forge recipe (see conda-forge/cantera-feedstock#27), plus (unrelated) removal of a few unnecessary declarations.
test_phase_order_surf_jacobian
, which was failing by a slim margin on ppc64le Linux buildsCMakeLists.txt
files -- see https://gitlab.kitware.com/cmake/community/-/wikis/doc/cmake/VariablesListsStrings for some help understanding the rather non-intuitive CMake syntax.RPATH
on macOSCXX
conda
layout when usingpackage_build
soln_newSolution
in clib demo to avoid undefined behavior if the input file is not found-isysroot
flags in package buildsMultiTransport::correctBinDiffCoeffs
which had no corresponding implementationfriend
declarations fromDenseMatrix
-- these methods either don't exist with the specified signature (invert
) or only access public members of theDenseMatrix
class (solve
) and don't need to be friends.Checklist
scons build
&scons test
) and unit tests address code coverage