Closed ocefpaf closed 6 years ago
I don't think it is a conda-build bug. We had an internal user report this also a little while back. I looked into it, but didn't see anything amiss. You may want to add
if %ERRORLEVEL% NEQ 0 exit 1
just below your conda-build call.
You may want to add
if %ERRORLEVEL% NEQ 0 exit 1
Thanks. Just send a commit to test that.
@msarahan same thing :-(
https://ci.appveyor.com/project/ocefpaf/udunits-2/build/1.0.17
That's annoying.
Here's what conda-build does:
outer try: https://github.com/conda/conda-build/blob/master/conda_build/build.py#L2120 inner subprocess call: https://github.com/conda/conda-build/blob/master/conda_build/build.py#L1417 except: https://github.com/conda/conda-build/blob/master/conda_build/build.py#L2249
There is nothing that catches that subprocess exception, FWICT. It absolutely should be erroring out, and it is, based on https://ci.appveyor.com/project/ocefpaf/udunits-2/build/1.0.17#L598
but somehow appveyor isn't reading that status correctly. Maybe try forcing appveyor to treat it as cmd.exe, not powershell?
Maybe try forcing appveyor to treat it as cmd.exe, not powershell?
That could be it. Not sure how to force that though. I'll read up the docs later. Thanks!
@ocefpaf & @msarahan, Hang on. I've made default_udunits2_xml_path()
static because the public function ut_get_xml_path()
should have been used by the udunits2
program, instead. This might obviate the problem.
This PR revives AppVeyor but, contrary to the green status, the build is failing with:
I have no idea why AppVeyor is reporting a green status when the last message is
subprocess.CalledProcessError: Command '['cmd.exe', '/c', 'bld.bat']' returned non-zero exit status 1.
@msarahan is this a bug in conda-build? Or am I using it wrong? This used to failed as expected some time ago :-/