Automating the generation of type stubs #1142
Conversation
| primal_bound: float = None | ||
| dual_bound: float = None | ||
| gap: float = None | ||
| primal_dual_integral: float = None |
There was a problem hiding this comment.
It has some problems with this statistics class, but this is going to be deprecated in favor of the JSON statistics
jonathanberthias
left a comment
jonathanberthias
left a comment
There was a problem hiding this comment.
This is a great script, I'm surprised at how effective it is at reproducing the current stubs!
However, I do think it is overfitting the current state and will break as soon as we start improving the stubs as it will try to revert any changes we make.
| lines.append(' @staticmethod') | ||
| lines.append(f' def {method}(*args, **kwargs): ...') | ||
| else: | ||
| lines.append(f' def {method}(self, *args, **kwargs): ...') |
There was a problem hiding this comment.
The script hardcodes the arguments to *args, **kwargs which won't work anymore once we start putting the exact arguments in the stubs.
There was a problem hiding this comment.
ah yes, this will need to be revisited once we have the arguments, but I'm fairly confident that the AI will manage to incorporate the new types. But, in any case, until we do add the types, I wouldn't be too worried about merging this.
@jonathanberthias when you have the time, can you please take a look? I deleted
scip.pyiand ran the script, generating thescip.pyithat's in this PR.I also wrapped
SCIPaddClique, and re-ran the script. It did add the stub toscip.pyi.