ENH: optimize.linprog: use callbacks for highs #19420
Conversation
Still needs a wrapper around Highs()
Co-authored-by: mckib2 <mckib2@users.noreply.github.com>
With fixes as suggested Co-authored-by: rgommers <rgommers@users.noreply.github.com>
Co-authored-by: rgommers <rgommers@users.noreply.github.com>
... outside of scipy at any rate
Remain 100% backwards compatible
a54cff5 to
6afbe49
Compare
|
@HaoZeke, any update here? It's probably getting a bit tight to get this into 1.14... |
Thanks for the ping, @h-vetinari. Also, I'm sorry, but this needs some work since the underlying PR got reverted (and is being reworked a bit upstream). I'll be ready with the previous (highs update PR) this weekend but I see from the release schedule (and the fact that reviews will be needed) that none of these will be in before rc2 at the earliest. |
|
I assume this won't be making it in this week... feel free to prove me wrong! |
|
@HaoZeke, are you still planning to update this PR? :) |
Probably not in the near future, I've a bit of personal work coming up and am traveling as well. I'd revise the ETA on this to be at least a couple of months. I'm also happy for anyone else to take over, it should be straightforward exposing the actual callbacks to highs, but since they now run at different times than the older variant there needs to be a lot of documentation for scipy users. |
|
@HaoZeke I'm guessing this will probably not make it into the |
|
Sorry, but this is a bit on the backburner for me at the moment, moving to a new position. I will try to get back in time for the next release. |
Do not attempt to review until #19255 is in. :)
Reference issue
Closes #15707 eventually.
What does this implement/fix?
Basically passes
callbackinformation in the same way as the rest oflinprog.Additional information
Draft until its ready. Might need more changes here as well. Depends on #19255.