Julian Hall
With the MPS file from JuMP, it's solved fine with/without presolve, but with the .lp file from JuMP, I reproduce the issue. Now I can investigate
It's a bug in presolve. Thanks for raising it
If a MIP is infeasible there's no dual ray
I'm not looking into this at present, as I'm preparing to leave for ISMP. Once I'm back (on 29/07) I'll have a lot of time for MIP work.
I don't know. I don't see why it should. In the case you've highlighted, HiGHS should report something more informative. I'll see that it does
Once #1359 is merged into `master`, HiGHS no longer logs anything or returns an error from `getDualRay` if the model is a MIP, since there will never be an invertible...
I restored the printing of the version number, but I can't see when. It's in [v1.6.0](https://github.com/ERGO-Code/HiGHS/blob/v1.6.0/src/io/HighsIO.cpp) Since 1.6.0 HiGHS - optionally - [logs the githash](src/io/HMPSIO.cpp). As for "resource files" when...
You make your point. How we are able to give the dll a version number is another question. I've no idea
The simplex solver is concerned about a risk of cycling, so has not made the final basis change required to get an optimal solution. I'll look at this in more...
Rather more useful for debugging is [issue-1561.c.txt](https://github.com/ERGO-Code/HiGHS/files/13859426/issue-1561.c.txt)