Closed ubfx closed 7 months ago
Thanks for the patch! Let me check w/ the CI if it fixes #368
still seeing:
2024-02-21T10:08:47.1051513Z [32;1m10:08:47[0m | Error was:
2024-02-21T10:08:47.1053469Z [32;1m10:08:47[0m | Command '['/root/conda-eda/conda-eda/workdir/conda-env/bin/patch', '--no-backup-if-mismatch', '--batch', '-Np1', '-i', '/tmp/tmpg6s06yi_/disable-cluster-flops-cmd.patch.native', '--binary', '--dry-run']' returned non-zero exit status 1.
Sorry, it seems like the files changed again since I last looked at it. I'm updating the patch now
Seems like this fixed it in my fork: https://github.com/ubfx/conda-eda/actions/runs/7987639828/job/21810507800#step:5:1480
@ubfx thanks for the quick turnaround, running the CI.
Looks like we miss an stdlib include?
2024-02-21T11:23:40.7739118Z [32;1m11:23:40[0m | /root/conda-eda/conda-eda/workdir/conda-env/conda-bld/openroad_1708513964138/work/src/dpl/src/Opendp.cpp:664:14: error: 'accumulate' is not a member of 'std'
2024-02-21T11:23:40.7740773Z [32;1m11:23:40[0m | 664 | = std::accumulate(grid_sites.begin(),
2024-02-21T11:23:40.7741574Z [32;1m11:23:40[0m | | ^~~~~~~~~~
2024-02-21T11:23:41.5824918Z [32;1m11:23:41[0m | make[2]: *** [src/dpl/CMakeFiles/dpl_lib.dir/build.make:76: src/dpl/CMakeFiles/dpl_lib.dir/src/Opendp.cpp.o] Error 1
2024-02-21T11:23:41.5826677Z [32;1m11:23:41[0m | make[1]: *** [CMakeFiles/Makefile2:2673: src/dpl/CMakeFiles/dpl_lib.dir/all] Error 2
2024-02-21T11:23:41.9383108Z [32;1m11:23:41[0m | make[1]: *** Waiting for unfinished jobs....
The ortools patch problem seems fixed, now we're seeing:
py37:
11:23:40 | /root/conda-eda/conda-eda/workdir/conda-env/conda-bld/openroad_1708513964138/work/src/dpl/include/dpl/Opendp.h: In member function 'int dpl::GridInfo::getSitesTotalHeight() const':
11:23:40 | /root/conda-eda/conda-eda/workdir/conda-env/conda-bld/openroad_1708513964138/work/src/dpl/include/dpl/Opendp.h:230:17: error: 'accumulate' is not a member of 'std'
py38 and py310:
/root/conda-eda/conda-eda/workdir/conda-env/conda-bld/openroad_1708513921126/work/src/gui/src/layoutTabs.cpp:113:11: error: 'qOverload' was not declared in this scope
11:22:30 | 113 | qOverload<const Selected&>(&LayoutViewer::addSelected),
The accumulate error I also saw on my runner, and can be fixed by adding #include <numeric>
. Not sure about the qOverload one because I haven't seen that before.
Let's see what the CI says ? https://github.com/hdl/conda-eda/actions/runs/7988449270/job/21813113707?pr=366 :)
still getting:
2024-02-21T11:59:28.9049370Z [32;1m11:59:28[0m | /root/conda-eda/conda-eda/workdir/conda-env/conda-bld/openroad_1708516137315/work/src/gui/src/layoutTabs.cpp:113:11: error: 'qOverload' was not declared in this scope
2024-02-21T11:59:28.9051137Z [32;1m11:59:28[0m | 113 | qOverload<const Selected&>(&LayoutViewer::addSelected),
for python > 3.7 . but openroad-linux-py37 seems to be passing! \o/ https://github.com/hdl/conda-eda/actions/runs/7988449270/job/21813113707?pr=366
maybe missing include or diverging CXX_STANDARD
version?
maybe missing include or diverging
CXX_STANDARD
version?
Seems like qOverload
was introduced in Qt 5.7, but the py37 build runs on 5.6.8. So there is probably a switch somewhere which disables qOverload for the py37 build, explaining why it doesn't see this error. For the 3.8 and 3.10 builds, building with newer Qt versions, we might have to #include <QtGlobal>
. I'll give that a try.
For the 3.8 and 3.10 builds, building with newer Qt versions, we might have to #include
. I'll give that a try.
should we pin the qt version to 5.6.8
?
Looking at the logs it looks like for py38 and py310 the version of qt that gets resolved is older!
2024-02-21T13:22:37.0321311Z [32;1m13:22:37[0m | - qt 5.6.3 h8bf5577_3
2024-02-21T13:22:50.5482877Z [32;1m13:22:50[0m | - qt 5.6.3 h8bf5577_3
while py37 resolves to:
2024-02-21T13:22:26.3106916Z [32;1m13:22:26[0m | - qt 5.9.7 h5867ecd_1
it looks like for py38 and py310 the version of qt that gets resolved is older!
Ah that's probably it! Not sure where I got those version numbers from earlier. 😄 By the way - I created a PR for the numeric include issue over at https://github.com/The-OpenROAD-Project/OpenROAD/pull/4696 so we will probably have to remove that patch again once that is merged.
all green!
@ubfx do you want me to wait for https://github.com/The-OpenROAD-Project/OpenROAD/pull/4696 to land before merging?
@ubfx do you want me to wait for The-OpenROAD-Project/OpenROAD#4696 to land before merging?
Up to you, I can also create a new PR on here as soon as it lands.
Let's unblock this then, and you (or I!) can follow up with a PR removing the patches once https://github.com/The-OpenROAD-Project/OpenROAD/pull/4696 lands.
Thanks a lot for working thru this together!
Currently, the OpenROAD build fails, because this patch can't be applied.
I actually had to do some more changes in my fork to get the build working again, but I'm not sure how much of that is because of my runner. So maybe this will do.
Fix https://github.com/hdl/conda-eda/issues/368