CEP 3 - Using the Mamba solver in conda. #2
Conversation
jezdez
changed the title
jezdez
changed the title
cep-3.md
Outdated
| Conda contains a solver registry in the `conda.resolve` module that | ||
| must be extended to use the plugin architecture specified in CEP 2. | ||
|
|
||
| Conda currently (4.10.x) supports three solvers: |
There was a problem hiding this comment.
One important differentiation as far as I understand mamba is also that the building of the SAT clauses is done in C++ and not in Python. This also saves a considerable amount of time and is different to the current interface in https://github.com/conda/conda/blob/248741a843e8ce9283fa94e6e4ec9c2fafeb76fd/conda/resolve.py#L44
We should still be able to provide a registry in the conda.resolve module, probably just on a different level than the current one.
There was a problem hiding this comment.
Yeah, this'll be part of the implementation of the CEP, I'll update the draft to include a note about the difference in API levels and a need to refactor and/or integrate appropriately. Thanks for raising this!
There was a problem hiding this comment.
@xhochy do you have any data on this? The claim(? - as far as I understand) that the SAT clause definition is somehow a bottleneck seems a bit weird to me 😅
Admittedly, I am not super knowledgeable on SAT solvers, but I would not expect this to take for than a couple milliseconds.
Usually, the bottleneck on dependency resolvers that work on remote sources is fetching the metadata necessary for the dependency resolution. So, it usually tends to be better to give more weight to the resolution algorithm instead of this kind of optimizations 😄
I haven't looked too much in all the possible backends, I am just raising the question.
There was a problem hiding this comment.
This meta issue has some useful links to understand the bottlenecks: conda/conda#7700
Co-authored-by: Jannis Leidel <jannis@leidel.info>
|
@conda-incubator/steering This PR falls under the Enhancement Proposal Approval policy of the conda governance policy, please vote and/or comment on this PR. This PR needs 60% of the Steering Council to vote yea to pass. To vote please leave Approve (yea) or Request Changes (nay) pull request reviews. If you would like changes to the current language please leave a comment (in the PR) or push to this branch. This vote will end on 2022-05-24. |
jezdez
changed the title
|
Hey all, this is just a reminder that another vote is happening in #23 Thank you! |
Voting resultsThis was a standard, non-timed-out vote.
This vote has reached quorum (17 + 0 = 17 which is at least 60% of 21). It has also passed since it recorded 18 "yes" votes and 0 "no" votes giving 18/18 which is greater than 60% of 18. |
This depends on #1 and is definitely the first pass only.