profile
viewpoint
If you are wondering where the data of this site comes from, please visit https://api.github.com/users/mathstuf/events. GitMemory does not store any data, but only uses NGINX to cache data for a period of time. The idea behind GitMemory is simply to give users a better reading experience.
Ben Boeckel mathstuf Kitware, Inc.

Kitware/ninja 37

Kitware branch of ninja for staging features not yet integrated upstream

Kitware/MetaIO 25

MetaImage is the text-based tagged file format for medical images that resulted. We have now extended that file format to support a variety of objects that occur in medicine such a tubes (for vessels, needles, etc.), blobs (for arbitrary shaped objects), cubes, spheres, etc. The complete library is known at MetaIO.

hack-rpi/Status-Board 14

Hackathon commit messages, announcements, mentor requests, and more, all in one place!

cplusplus/modules-ecosystem-tr 11

The ISO C++ Modules Ecosystem Technical Report (METeR)

mathstuf/abagames-gunroar 7

Modernization of gunroar (http://www.asahi-net.or.jp/~cs8k-cyu/windows/gr_e.html)

mathstuf/abagames-torus-trooper 5

Modernization of torus trooper (http://www.asahi-net.or.jp/~cs8k-cyu/windows/tt_e.html)

Kitware/paraview-docs 3

ParaView documentation

mathstuf/abagames-tumiki-fighters 2

Modernization of tumiki fighters (http://www.asahi-net.or.jp/~cs8k-cyu/windows/tf_e.html)

CommitCommentEvent

delete branch mathstuf/qhull

delete branch : dont-force-install-properties

delete time in 9 days

startedtaiki-e/cargo-llvm-cov

started time in 11 days

issue closednumpy/numpy

NumPy selects Accelerate.framework preferentially over netlib

I'm building NumPy 1.21.1 from source and aiming to build against netlib's BLAS and LAPACK (3.10.0). However, even giving the environment variables NPY_BLAS_ORDER=blas and NPY_LAPACK_ORDER=lapack ends up linking against Accelerate.framework instead. The build output says it found the blas and lapack I intend to use, but the resulting binaries are still linked against Accelerate.framework. What other controls are available?

Relevant output from the build (with build/install prefixes replaced):

blas_opt_info:
…/numpy/distutils/system_info.py:2026: UserWarning: 
    Optimized (vendor) Blas libraries are not found.
    Falls back to netlib Blas library which has worse performance.
    A better performance should be easily gained by switching
    Blas library.
  if self._calc_info(blas):
blas_info:
customize UnixCCompiler
C compiler: gcc -Wno-unused-result -Wsign-compare -Wunreachable-code -DNDEBUG -g -fwrapv -O3 -Wall -fPIC -mmacosx-version-min=10.13 -mmacosx-version-min=10.13 -fPIC -mmacosx-version-min=10.13 -mmacosx-version-min=10.13 -fPIC -mmacosx-versi
on-min=10.13

creating /var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr/T/tmp0v6k1o2h/var
creating /var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr/T/tmp0v6k1o2h/var/folders
creating /var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr/T/tmp0v6k1o2h/var/folders/zk
creating /var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr/T/tmp0v6k1o2h/var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr
creating /var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr/T/tmp0v6k1o2h/var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr/T
creating /var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr/T/tmp0v6k1o2h/var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr/T/tmp0v6k1o2h
compile options: '-I<PREFIX>/lib -I/usr/local/include -I<PREFIX>/include -c'
gcc: /var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr/T/tmp0v6k1o2h/source.c
/var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr/T/tmp0v6k1o2h/source.c:1:10: fatal error: 'cblas.h' file not found
#include <cblas.h>
         ^~~~~~~~~
1 error generated.
  FOUND:
    libraries = ['blas', 'blas']
    library_dirs = ['<PREFIX>/lib']
    include_dirs = ['<PREFIX>/lib', '/usr/local/include', '<PREFIX>/include']
    runtime_library_dirs = ['<PREFIX>/lib']

  FOUND:
    define_macros = [('NO_ATLAS_INFO', 1)]
    libraries = ['blas', 'blas']
    library_dirs = ['<PREFIX>/lib']
    include_dirs = ['<PREFIX>/lib', '/usr/local/include', '<PREFIX>/include']
    runtime_library_dirs = ['<PREFIX>/lib']

non-existing path in 'numpy/distutils': 'site.cfg'
lapack_opt_info:
lapack_info:
  FOUND:
    libraries = ['lapack', 'lapack']
    library_dirs = ['<PREFIX>/lib']
    language = f77
    runtime_library_dirs = ['<PREFIX>/lib']

  FOUND:
    libraries = ['lapack', 'lapack', 'blas', 'blas']
    library_dirs = ['<PREFIX>/lib']
    language = f77
    runtime_library_dirs = ['<PREFIX>/lib']
    define_macros = [('NO_ATLAS_INFO', 1)]
    include_dirs = ['<PREFIX>/lib', '/usr/local/include', '<PREFIX>/include']

Link information of libraries installed:

% otool -L <PREFIX>/lib/python3.9/site-packages/numpy/linalg/_umath_linalg.cpython-39-darwin.so
<PREFIX>/lib/python3.9/site-packages/numpy/linalg/_umath_linalg.cpython-39-darwin.so:
        /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libLAPACK.dylib (compatibility version 1.0.0, current version 1.0.0)
        /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib (compatibility version 1.0.0, current version 1.0.0)
        /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 1292.60.1)

closed time in 13 days

mathstuf

issue commentnumpy/numpy

NumPy selects Accelerate.framework preferentially over netlib

Oh, the flags come from our patchups to help gfortran find the libSystem.tbd for linking now that the file isn't in /usr/lib anymore.

I suppose the thing to do here would be to prefer listing the .dylib on the link line directly rather than relying on -L flags to resolve to the intended name via -l. Sorry for the noise.

mathstuf

comment created time in 13 days

issue commentnumpy/numpy

NumPy selects Accelerate.framework preferentially over netlib

OK, after digging in, it seems the problem is that -L/path/to/MacOSX.sdk/usr/lib is passed before my blas build's. This makes the linker find the libblas.tbd file in the SDK and ends up using Accelerate.framework that way. Removing this manually from the link line works. Now just how to figure out to make the build system not do that.

mathstuf

comment created time in 13 days

issue commentnumpy/numpy

NumPy selects Accelerate.framework preferentially over netlib

This is a link-time thing AFAICT. There is an explicit mention of Accelerate.framework in the LC_LOAD_DYLIB command for the module, so there's no rpath or runtime lookup going on here.

mathstuf

comment created time in 13 days

issue openednumpy/numpy

NumPy selects Accelerate.framework preferentially over netlib

I'm building NumPy 1.21.1 from source and aiming to build against netlib's BLAS and LAPACK (3.10.0). However, even giving the environment variables NPY_BLAS_ORDER=blas and NPY_LAPACK_ORDER=lapack ends up linking against Accelerate.framework instead. The build output says it found the blas and lapack I intend to use, but the resulting binaries are still linked against Accelerate.framework. What other controls are available?

Relevant output from the build (with build/install prefixes replaced):

blas_opt_info:
…/numpy/distutils/system_info.py:2026: UserWarning: 
    Optimized (vendor) Blas libraries are not found.
    Falls back to netlib Blas library which has worse performance.
    A better performance should be easily gained by switching
    Blas library.
  if self._calc_info(blas):
blas_info:
customize UnixCCompiler
C compiler: gcc -Wno-unused-result -Wsign-compare -Wunreachable-code -DNDEBUG -g -fwrapv -O3 -Wall -fPIC -mmacosx-version-min=10.13 -mmacosx-version-min=10.13 -fPIC -mmacosx-version-min=10.13 -mmacosx-version-min=10.13 -fPIC -mmacosx-versi
on-min=10.13

creating /var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr/T/tmp0v6k1o2h/var
creating /var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr/T/tmp0v6k1o2h/var/folders
creating /var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr/T/tmp0v6k1o2h/var/folders/zk
creating /var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr/T/tmp0v6k1o2h/var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr
creating /var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr/T/tmp0v6k1o2h/var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr/T
creating /var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr/T/tmp0v6k1o2h/var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr/T/tmp0v6k1o2h
compile options: '-I<PREFIX>/lib -I/usr/local/include -I<PREFIX>/include -c'
gcc: /var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr/T/tmp0v6k1o2h/source.c
/var/folders/zk/2_2hy13546dc7wdq4tslst3r0000gr/T/tmp0v6k1o2h/source.c:1:10: fatal error: 'cblas.h' file not found
#include <cblas.h>
         ^~~~~~~~~
1 error generated.
  FOUND:
    libraries = ['blas', 'blas']
    library_dirs = ['<PREFIX>/lib']
    include_dirs = ['<PREFIX>/lib', '/usr/local/include', '<PREFIX>/include']
    runtime_library_dirs = ['<PREFIX>/lib']

  FOUND:
    define_macros = [('NO_ATLAS_INFO', 1)]
    libraries = ['blas', 'blas']
    library_dirs = ['<PREFIX>/lib']
    include_dirs = ['<PREFIX>/lib', '/usr/local/include', '<PREFIX>/include']
    runtime_library_dirs = ['<PREFIX>/lib']

non-existing path in 'numpy/distutils': 'site.cfg'
lapack_opt_info:
lapack_info:
  FOUND:
    libraries = ['lapack', 'lapack']
    library_dirs = ['<PREFIX>/lib']
    language = f77
    runtime_library_dirs = ['<PREFIX>/lib']

  FOUND:
    libraries = ['lapack', 'lapack', 'blas', 'blas']
    library_dirs = ['<PREFIX>/lib']
    language = f77
    runtime_library_dirs = ['<PREFIX>/lib']
    define_macros = [('NO_ATLAS_INFO', 1)]
    include_dirs = ['<PREFIX>/lib', '/usr/local/include', '<PREFIX>/include']

Link information of libraries installed:

% otool -L <PREFIX>/lib/python3.9/site-packages/numpy/linalg/_umath_linalg.cpython-39-darwin.so
<PREFIX>/lib/python3.9/site-packages/numpy/linalg/_umath_linalg.cpython-39-darwin.so:
        /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libLAPACK.dylib (compatibility version 1.0.0, current version 1.0.0)
        /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib (compatibility version 1.0.0, current version 1.0.0)
        /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 1292.60.1)

created time in 14 days

pull request commentpython/cpython

bpo-38728: Update PC/pyconfig.h to allow disabling pragma based auto-linking

Seems this will miss 3.10 as well.

jcfr

comment created time in 16 days

issue commentrust-lang/rust

Provide an API to extract fields from Command builder

Does not offer a way to get an environment variable by name (get_env). I figure this can be added later if anyone really wants it. I think the motivation for this is weak, though. Also, the API could be a little awkward (return a Option<Option<&OsStr>>?).

Not that it affects this PR, but something like this would be fine I think:

struct ProcessEnvSetting<'a> {
  Inherit,
  Unset,
  Set(&'a OsStr),
}

impl<'a> ProcessEnvSetting<'a> {
  // Convenience method.
  fn value_in_child(&self) -> Option<Cow<'a, OsStr>> {
    // query current environment if `Inherit`
  }
}

env_clear could then set an internal flag to return Unset rather than Inherit for any variable not explicitly set.

euclio

comment created time in 18 days

issue openedmozilla/sccache

Support wrapping clang-tidy as well

clang-tidy is used in CI and being able to have its linting be cached would help immensely with its turnaround times. Currently, the compilation can be cached, but it'd be nice to have tidy output be cached as well.

Of note for the implementation:

  • no -o object artifact applies
  • implicit .clang-tidy files need to be considered in the hash since these affect what flags are being used internally

created time in 20 days

issue commentAcademySoftwareFoundation/openvdb

[BUILD] Please provide an `openvdb-config.cmake` file instead of FindOpenVDB

I looked at it before filing the issue, but it didn't fit into the time I had available then. I can certainly help review if someone gets to it sooner, but I don't know when I'll get the time to work on it as of right now.

mathstuf

comment created time in 20 days

issue openedpimutils/khal

Documentation mentions an unworking command

Reported here: https://bugzilla.redhat.com/show_bug.cgi?id=1998868

The command in doc/source/usage.rst mentions a command khal new 25.10. 18:00 24:00 Another Event :: with Alice and Bob which ends up with a backtrace;

Traceback (most recent call last):
  File "/usr/bin/khal", line 33, in <module>
    sys.exit(load_entry_point('khal==0.9.10', 'console_scripts', 'khal')())
  File "/usr/lib/python3.9/site-packages/click/core.py", line 829, in __call__
    return self.main(*args, **kwargs)
  File "/usr/lib/python3.9/site-packages/click/core.py", line 782, in main
    rv = self.invoke(ctx)
  File "/usr/lib/python3.9/site-packages/click/core.py", line 1259, in invoke
    return _process_result(sub_ctx.command.invoke(sub_ctx))
  File "/usr/lib/python3.9/site-packages/click/core.py", line 1066, in invoke
    return ctx.invoke(self.callback, **ctx.params)
  File "/usr/lib/python3.9/site-packages/click/core.py", line 610, in invoke
    return callback(*args, **kwargs)
  File "/usr/lib/python3.9/site-packages/click/decorators.py", line 21, in new_func
    return f(get_current_context(), *args, **kwargs)
  File "/usr/lib/python3.9/site-packages/khal/cli.py", line 386, in new
    new_func(
  File "/usr/lib/python3.9/site-packages/khal/controllers.py", line 339, in new_from_string
    info = utils.eventinfofstr(info, conf['locale'], adjust_reasonably=True, localize=False)
  File "/usr/lib/python3.9/site-packages/khal/utils.py", line 521, in eventinfofstr
    raise ValueError('Could not parse `{}`'.format(info_string))
ValueError: Could not parse `25.10. 18:00 24:00 Another Event

created time in 22 days

PR opened qhull/qhull

Dont force install properties

Various distributions may want to control these installation properties themselves for a variety of reasons (e.g., removing rpath settings, preferring @rpath/ on macOS, or whatever). Qhull's manual setting of these properties interferes with this ability. Instead, just default the variables if they're not already set through the cache (usually passing -D on when configuring).

+16 -27

0 comment

1 changed file

pr created time in 25 days

create barnchmathstuf/qhull

branch : dont-force-install-properties

created branch time in 25 days

issue commentspack/spack

Installation issue: Paraview fails on cmake.

That sounds reasonable to me.

navdeeprana

comment created time in 25 days

issue commentspack/spack

Installation issue: Paraview fails on cmake.

ParaView will start to ignore +kits in a ~shared build, so I would recommend backporting this hunk for 5.9.1.

navdeeprana

comment created time in a month

issue commentspack/spack

Installation issue: Paraview fails on cmake.

Taking a look at it, the change is just not eligible for 5.9 anyways (and I'd caution against a backport). I'd just deny the configuration of +static+kits+python3 up front for paraview@:5.10.

navdeeprana

comment created time in a month

issue commentspack/spack

Installation issue: Paraview fails on cmake.

I have a cycle involving some of the same bits here. I'll work on it for 5.10, but a backport to 5.9 can be done for Spack at least (I don't think we'll have a 5.9.2).

navdeeprana

comment created time in a month

issue commentspack/spack

Installation issue: Paraview fails on cmake.

Hmm. +python3 is definitely involved there, though I don't see what would cause this based on master right now. I'd need to dig in later.

navdeeprana

comment created time in a month

issue openedAcademySoftwareFoundation/openvdb

[BUILD] Please provide an `openvdb-config.cmake` file instead of FindOpenVDB

Environment

Operating System: any Version / Commit SHA: master CMake Version: any Compiler: irrelevant

Describe the problem

OpenVDB provides its own FindOpenVDB file. Instead, it should provide an openvdb-config.cmake (or OpenVDBConfig.cmake) file which just provides the information directly. OpenVDB's install tree already knows where everything is and having to re-find it is unnecessary work.

Please just export the targets and include them from the config file (which can do the necessary component work to only include the targets that are requested).

Additionally, there are issues related to policy scope. Performing cmake_minimum_required in an included file leaks the policy settings from it to the caller. Such calls should instead be:

cmake_policy(PUSH)
cmake_policy(VERSION 3.12)

# Rest of the script.

cmake_policy(POP)

so that the settings don't "leak" out.

created time in a month

delete branch mathstuf/openvdb

delete branch : sharpening_filter

delete time in a month

delete branch mathstuf/openvdb

delete branch : feature/sharpen

delete time in a month

delete branch mathstuf/openvdb

delete branch : gh-pages

delete time in a month

delete branch mathstuf/openvdb

delete branch : feature/pybind

delete time in a month

delete branch mathstuf/openvdb

delete branch : feature/nanovdb

delete time in a month

delete branch mathstuf/openvdb

delete branch : feature/ci2022

delete time in a month

fork mathstuf/openvdb

OpenVDB - Sparse volume data structure and tools

http://www.openvdb.org/

fork in a month

delete branch mathstuf/hdf5

delete branch : camke-variable-deref-issues

delete time in a month

PR opened HDFGroup/hdf5

cmake: simplify try_run result comparisons

Instead of looking at MATCHES 0 which runs a regex and returns a false positive for any result such as 10, check if they are equal to 0. Also quote 0 to avoid looking it up as a variable (which was left behind prior to the fix to use ${RETURN_VAR} in the c_run macro).

+5 -5

0 comment

1 changed file

pr created time in a month