Repository navigation
Add skimage.morphology.thin - #27
Conversation
grlee77
added
feature request
ajschmidt8
left a comment
ajschmidt8
left a comment
There was a problem hiding this comment.
Approving ops-codeowner file changes
|
rerun tests |
|
From CI: Maybe there is still something to be worked out here? |
Yeah, this is odd. As far as I can tell the function is present in the |
|
Moving to 21.08 as 21.06 is being shipped. Hopefully we can get this in the new release |
|
I need to review the CI a bit more (maybe I am misunderstanding something), but why do I see in those logs in the sections on testing: why is the cucim listed as above from rapidsai-nightly and not from some locally built version corresponding to this PR? |
|
Would wait for a bit before digging in. Suspect we are missing things like PR ( #53 ) |
|
Looks like the
|
Thanks @ajschmidt8 for the pointer! I would look at this and fix this today. |
We need both version strings -- zero-padded(`08`) and normal integer `8` in c++ code. ```c++ #define CUCIM_VERSION_MAJOR 21 #define CUCIM_VERSION_MINOR 08 #define CUCIM_VERSION_PATCH 00 ``` For zero-padded VERSION, it uses XSTR macro: ```c++ #define STR(x) #x #define XSTR(x) STR(x) ``` For integer version (uint32_t), it uses `std::atoi` with XSTR macro: Example, ```c++ static_cast<uint32_t>(std::atoi(XSTR(CUCIM_VERSION_MINOR))) ``` Fix a build issue in #27. Authors: - Gigon Bae (https://github.com/gigony) Approvers: - https://github.com/jakirkham URL: #59
|
rerun tests |
|
CI is back to the error shown in comment ( #27 (comment) ) |
|
Looking through the logs a bit more, cucim for this commit builds successfully with conda-build and ends up at: The following command SHOULD be installing that version to use when running gpuci_conda_retry install -y -c /workspace/ci/artifacts/cucim/cpu/.conda-bld -c /opt/conda/envs/rapids/conda-bld -c rapidsai-nightly 'rapids-build-env=21.08.*' libcucim cucimHowever, there is an error about package inconsistency that is causing a different version of cucim to be retrieved from The above excerpt is from the following log: |
|
Is there a simple way to constrain |
You could try splitting the following line and using i.e.: gpuci_conda_retry install -y \
-c /workspace/ci/artifacts/cucim/cpu/.conda-bld \
-c /opt/conda/envs/rapids/conda-bld \
-c rapidsai-nightly \
'rapids-build-env=21.08.*' \
libcucim \
cuciminto gpuci_conda_retry install -y \
-c rapidsai-nightly \
'rapids-build-env=21.08.*'
gpuci_conda_retry install -y \
--override-channels \
-c /workspace/ci/artifacts/cucim/cpu/.conda-bld \
-c /opt/conda/envs/rapids/conda-bld \
libcucim \
cucimThough I'm not sure how/if this will work if |
|
Splitting the command forced the local version to be used so that it now errors out with more messages about the conflicts. The reported conflicts look compiler/C library related ( The following claiming there is a CUDA mismatch looks odd to me as well (it appears both the detected and expected are actually 11.2?): Details |
I'm not great with compiler stuff, but I'm wondering if @jakirkham has any thoughts or input on this. I checked the package versions from the integration repo to see if they match up with both the |
#62 has the same issue with this. @ajschmidt8
Detailsgbae@GIGONY-DESKTOP:~/repo/cucim$ conda create -n rapids-env -c rapidsai-nightly rapids-build-env
Collecting package metadata (current_repodata.json): done
Solving environment: failed with repodata from current_repodata.json, will retry with next repodata source.
Collecting package metadata (repodata.json): done
Solving environment: |
Found conflicts! Looking for incompatible packages.
This can take several minutes. Press CTRL-C to abort.
failed -
UnsatisfiableError: The following specifications were found to be incompatible with your system:
- feature:/linux-64::__glibc==2.31=0
- rapids-build-env -> ucx -> __glibc[version='>=2.17|>=2.17,<3.0.a0']
Your installed version is: 2.31
gbae@GIGONY-DESKTOP:~/repo/cucim$ conda info
active environment : None
shell level : 0
user config file : /home/gbae/.condarc
populated config files : /home/gbae/.condarc
conda version : 4.10.1
conda-build version : 3.21.4
python version : 3.8.5.final.0
virtual packages : __linux=5.10.16.3=0
__glibc=2.31=0
__unix=0=0
__archspec=1=x86_64
base environment : /home/gbae/miniconda3 (writable)
conda av data dir : /home/gbae/miniconda3/etc/conda
conda av metadata url : https://repo.anaconda.com/pkgs/main
channel URLs : https://repo.anaconda.com/pkgs/main/linux-64
https://repo.anaconda.com/pkgs/main/noarch
https://repo.anaconda.com/pkgs/r/linux-64
https://repo.anaconda.com/pkgs/r/noarch
package cache : /home/gbae/miniconda3/pkgs
/home/gbae/.conda/pkgs
envs directories : /home/gbae/miniconda3/envs
/home/gbae/.conda/envs
platform : linux-64
user-agent : conda/4.10.1 requests/2.25.1 CPython/3.8.5 Linux/5.10.16.3-microsoft-standard-WSL2 ubuntu/20.04.1 glibc/2.31
UID:GID : 1000:1000
netrc file : None
offline mode : False
|
|
Can you please try specifying |
It doesn't help :( gbae@GIGONY-DESKTOP:~/repo/cucim$ conda create -n rapids-env -c rapidsai-nightly rapids-build-env cudatoolkit=11.0
Collecting package metadata (current_repodata.json): done
Solving environment: failed with repodata from current_repodata.json, will retry with next repodata source.
Collecting package metadata (repodata.json): done
Solving environment: -
Found conflicts! Looking for incompatible packages.
This can take several minutes. Press CTRL-C to abort.
Examining rapids-build-env: 50%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ | 3/6 [00:00<00:00, 5790.57it- ^failed
CondaError: KeyboardInterruptEven this command is hanging gbae@GIGONY-DESKTOP:~/repo/cucim$ conda create -n rapids-env -c gpuci -c rapidsai-nightly -c rapidsai -c nvidia -c pytorch -c conda-forge rapids-build-env cudatoolkit=11.0
Collecting package metadata (current_repodata.json): done
Solving environment: failed with repodata from current_repodata.json, will retry with next repodata source.
Collecting package metadata (repodata.json): done
Solving environment: \
|
We need both version strings -- zero-padded(`08`) and normal integer `8` in c++ code. ```c++ #define CUCIM_VERSION_MAJOR 21 #define CUCIM_VERSION_MINOR 08 #define CUCIM_VERSION_PATCH 00 ``` For zero-padded VERSION, it uses XSTR macro: ```c++ #define STR(x) #x #define XSTR(x) STR(x) ``` For integer version (uint32_t), it uses `std::atoi` with XSTR macro: Example, ```c++ static_cast<uint32_t>(std::atoi(XSTR(CUCIM_VERSION_MINOR))) ``` Fix a build issue in rapidsai#27. Authors: - Gigon Bae (https://github.com/gigony) Approvers: - https://github.com/jakirkham URL: rapidsai#59
|
Okay, the conda conflicts seem to have been resolved elsewhere and this is working now! |
|
Yeah I suspect we were just running into the same Arrow issues we saw elsewhere in RAPIDS. Now that those have been resolved this has cleared up |
|
@gpucibot merge |
This PR adds one additional function for thinning binary images as shown here. This particular algorithm is 2D only. There is a second skeletonization algorithm in scikit-image with 3D support, but that one is currently Cython based and take more work to adapt.