DEVOPS-135 Fix capnproto object/code mismatch in manylinux build #1030
Conversation
|
By analyzing the blame information on this pull request, we identified @akhilaananthram, @scottpurdy and @rcrowder to be potential reviewers |
vitaly-krugl
force-pushed
the
devops135-fix-futex-hang
branch
5 times, most recently
from
9576690 to
de44526
Compare
…er to avoid conflict with capnproto methods compiled into pycapnp, possibly compiled with a different compiler/version and with different compiler/linker flags. Instead, we rely on dynamic linking of the necessary symbols from capnproto in pycapnp.
vitaly-krugl
force-pushed
the
devops135-fix-futex-hang
branch
20 times, most recently
from
447eb41 to
3e82745
Compare
|
|
||
| # | ||
| # tests_all just calls other targets | ||
| # | ||
| # TODO This doesn't seem to have any effect; it's probably because the DEPENDS |
There was a problem hiding this comment.
@scottpurdy: running of the tests from here is (and has been) dysfunctional. We actually run the tests from our higher-level build script after make completes. Do we really want to run tests during make? If not, we should remove this and the underlying custom targets, such as tests_cpp_region. What do you think? Then, I can file the appropriate issue and remove this TODO comment.
…ions DLL in RTLD_GLOBAL mode, which enables resultion of capnproto references when nupic.bidnings extensions are loaded.
…against the "combined" nupic_core static library, which is considered an output artifact, instead of nupic_core_solo static lib, which is only an intermediate step.
…termediate artifact and not intended to be an output product.
…ifying system libraries. It's a static library and shouldn't need additional linking information.
nupic_core_solo is the only consumer of the custom command's outputs.
…arget name
${CAPNP_STATIC_LIB_TARGET} and containing all capnproto library objects.
vitaly-krugl
force-pushed
the
devops135-fix-futex-hang
branch
2 times, most recently
from
e18dba5 to
7e2f4c0
Compare
|
@scottpurdy, the hick-up with cmake 2.8.7 (on travis's default linux environment) has been resolved; this PR's build should go back to green now. Fire away! |
|
@scottpurdy: parallelized build also works now; I tested with |
…cLib depend directly on external project Apr1StaticLib instead of its library wrapper ${APR1_STATIC_LIB_TARGET}; the latter was incorrectly interperting the dependency as another external project instead of library; but worked correctly on cmake 2.8.12.
Try building windows AprUtil1StaticLib with empty string -DAPR_LIBRARIES value.
vitaly-krugl
force-pushed
the
devops135-fix-futex-hang
branch
from
7e2f4c0 to
f800e70
Compare
| @@ -171,7 +149,7 @@ import_array(); | |||
| x_min, x_max, 1, 65535); | |||
| return y.forPython(); | |||
| } | |||
| */ | |||
| */ | |||
There was a problem hiding this comment.
Was this file missing the close for a multiline comment? I don't see an opening added.
There was a problem hiding this comment.
@scottpurdy, regarding
Was this file missing the close for a multiline comment? I don't see an opening added
No, nothing changed in that regard. When I made other changes in this file, the editor removed trailing spaces, which is what you're seeing on this line.
There was a problem hiding this comment.
Oh wow missed that, my bad. Make the old school diff is better 😉
vitaly-krugl
force-pushed
the
devops135-fix-futex-hang
branch
from
c1ce71b to
a1f6cda
Compare
vitaly-krugl
force-pushed
the
devops135-fix-futex-hang
branch
from
a1f6cda to
fd33070
Compare
|
Do you think it would be cleaner to put the misc. |
|
Other than |
|
@scottpurdy, regarding
It seems alright to me as is, but I must admit I might not have a good feel for cmake project structure - this is the only one I've every worked on. What to do? |
|
Offline, @scottpurdy agreed to merge as is. |
Fixes #1013
Forced runtime linking of nupic.bindings extensions against capnproto symbols in the pycapnp extension
Fixes #1034
Reorganized build to link nupic_core static library tests against the "combined" nupic_core static library, which is considered an output artifact, instead of nupic_core_solo static lib, which is only an intermediate step.
Along the way, cleaned up dependencies and refactored merging of static libs.
Afterthought: it might have been better as separate PRs, but I was taking a deep dive in the understanding of the build files, and it was more efficient to make fixes along the way that supported upstream development, rather than coming back for another pass.