Skip to content

Comments

Solver changes without eigen#1158

Merged
phkahler merged 9 commits intosolvespace:masterfrom
rpavlik:solver-changes-without-eigen
Dec 30, 2021
Merged

Solver changes without eigen#1158
phkahler merged 9 commits intosolvespace:masterfrom
rpavlik:solver-changes-without-eigen

Conversation

@rpavlik
Copy link
Contributor

@rpavlik rpavlik commented Dec 23, 2021

This is just the solver changes from #721, skipping the final commits that add Eigen, and fixed so that every commit successfully builds. I do not fully understand the changes, I just can rebase them sensibly. @Evil-Spirit is the real author here.

I'll put the Eigen stuff in a separate PR.

This does run.

With sanitizers enabled, looks like no new leaks thru here.

Copy link
Member

@phkahler phkahler left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I can't speak to the more complex changes, but nothing "looks wrong".

@rpavlik
Copy link
Contributor Author

rpavlik commented Dec 24, 2021

With this plus the "no sparse expr" pr, I can actually say I understand the Jacobian code moderately ok now, at least the part about computing and evaluating it. Haven't looked at the other methods in the file enough to learn those. For a fun time, look at branches I pushed to my fork recently named things like "another way to cause memory corruption" 😅. I should probably script running over the test case repo with sanitizers on to make sure it works on more than just the normal test suite and the 2 models I tested on .

rpavlik and others added 9 commits December 26, 2021 21:15
1. We are making FoldConstants first, so we are copying less amount of data in DeepCopyWithParamsAsPointers.
2. Since we already perform DeepCopyWithParamsAsPointers, PartialWrt already produces params as pointers
(Split out from earlier "Eigen library integration" commit)
Split from previous large 'Eigen library integration' commit.
@rpavlik rpavlik force-pushed the solver-changes-without-eigen branch from ee28073 to 7b03bd6 Compare December 27, 2021 03:16
@rpavlik
Copy link
Contributor Author

rpavlik commented Dec 27, 2021

OK this has been rebased, now builds on #1155 to simplify the history.

@phkahler phkahler merged commit e528fab into solvespace:master Dec 30, 2021
This was referenced Jan 6, 2022
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants