Closed odow closed 2 years ago
@lgottwald @jajhall looks like there are some segfaults in presolve somewhere: https://github.com/jump-dev/HiGHS.jl/runs/5117936095?check_suite_focus=true
signal (11): Segmentation fault
[152](https://github.com/jump-dev/HiGHS.jl/runs/5117936095?check_suite_focus=true#step:6:152)
in expression starting at /home/runner/work/HiGHS.jl/HiGHS.jl/test/MOI_wrapper.jl:243
[153](https://github.com/jump-dev/HiGHS.jl/runs/5117936095?check_suite_focus=true#step:6:153)
_ZN8presolve9HPresolve4linkEi at /home/runner/.julia/artifacts/b35ce16fd0665f735f8c25e9b0c5ca2209d78208/lib/libhighs.so (unknown line)
[154](https://github.com/jump-dev/HiGHS.jl/runs/5117936095?check_suite_focus=true#step:6:154)
_ZN8presolve9HPresolve7fromCSCERKSt6vectorIdSaIdEERKS1_IiSaIiEES9_ at /home/runner/.julia/artifacts/b35ce16fd0665f735f8c25e9b0c5ca2209d78208/lib/libhighs.so (unknown line)
[155](https://github.com/jump-dev/HiGHS.jl/runs/5117936095?check_suite_focus=true#step:6:155)
_ZN8presolve9HPresolve8setInputER7HighsLpRK12HighsOptionsP10HighsTimer at /home/runner/.julia/artifacts/b35ce16fd0665f735f8c25e9b0c5ca2209d78208/lib/libhighs.so (unknown line)
[156](https://github.com/jump-dev/HiGHS.jl/runs/5117936095?check_suite_focus=true#step:6:156)
_ZN17PresolveComponent3runEv at /home/runner/.julia/artifacts/b35ce16fd0665f735f8c25e9b0c5ca2209d78208/lib/libhighs.so (unknown line)
[157](https://github.com/jump-dev/HiGHS.jl/runs/5117936095?check_suite_focus=true#step:6:157)
_ZN5Highs11runPresolveEv at /home/runner/.julia/artifacts/b35ce16fd0665f735f8c25e9b0c5ca2209d78208/lib/libhighs.so (unknown line)
[158](https://github.com/jump-dev/HiGHS.jl/runs/5117936095?check_suite_focus=true#step:6:158)
_ZN5Highs3runEv at /home/runner/.julia/artifacts/b35ce16fd0665f735f8c25e9b0c5ca2209d78208/lib/libhighs.so (unknown line)
[159](https://github.com/jump-dev/HiGHS.jl/runs/5117936095?check_suite_focus=true#step:6:159)
Highs_run at /home/runner/work/HiGHS.jl/HiGHS.jl/src/gen/libhighs.jl:47 [inlined]
[160](https://github.com/jump-dev/HiGHS.jl/runs/5117936095?check_suite_focus=true#step:6:160)
optimize! at /home/runner/work/HiGHS.jl/HiGHS.jl/src/MOI_wrapper.jl:1562
I'll work to exclude the failing tests and then build some reproducible examples.
Bummer! Good job I've not advertised v1.2.0 directly to users. Rome wasn't built in a day...
This seems to be the underlying issue: https://github.com/ERGO-Code/HiGHS/issues/717
A lot of JuMP tests (and users) have a work-flow that looks like add variables -> constraints -> more variables -> more constraints.
Thanks. I'll create a HiGHS unit test for this and get it fixed today.
Merging #81 (1d89901) into master (221489a) will decrease coverage by
1.03%
. The diff coverage is89.58%
.
@@ Coverage Diff @@
## master #81 +/- ##
==========================================
- Coverage 82.65% 81.61% -1.04%
==========================================
Files 2 3 +1
Lines 1199 1213 +14
==========================================
- Hits 991 990 -1
- Misses 208 223 +15
Impacted Files | Coverage Δ | |
---|---|---|
src/gen/libhighs.jl | 34.74% <ø> (ø) |
|
src/MOI_wrapper.jl | 92.91% <88.88%> (-1.48%) |
:arrow_down: |
src/HiGHS.jl | 100.00% <100.00%> (ø) |
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update 221489a...1d89901. Read the comment docs.
@jajhall I updated the binaries to https://github.com/ERGO-Code/HiGHS/commit/f73254cfaa532139075fb174aa0c2e06e12cee34 and things look great.
We're passing all the JuMP tests on Windows, Mac, and Linux, on 64-bit and 32-bit, and we're testing the simplex and IPM solvers with and without presolve.
How about MIP?
Yes, all the MIP problems are passing as well
Actually I found one issue, but it's a compilation problem, not a correctness blocker: https://github.com/ERGO-Code/HiGHS/issues/734
Closes https://github.com/jump-dev/HiGHS.jl/issues/70 Closes https://github.com/jump-dev/HiGHS.jl/issues/73 Closes https://github.com/jump-dev/HiGHS.jl/issues/80