[C++] Fix GLPK bound type + numerical issues

hungpham2511 / toppra

robotic motion planning library

https://hungpham2511.github.io/toppra/index.html

MIT License

625 stars 170 forks source link

[C++] Fix GLPK bound type + numerical issues #111

Closed jmirabel closed 4 years ago

jmirabel commented 4 years ago

Fixes #

Changes in this PRs:

Checklists:

[x] Update CHANGELOG.md with a single line describing this PR

jmirabel commented 4 years ago

I use https://cli.github.com/ and, as I discover it now, it does not follow PR templates...

hungpham2511 commented 4 years ago

No worry. Thanks for the PR. Could you help to add a line to HISTORY.md? Then feel free to merge whenever you are happy.

jmirabel commented 4 years ago

Do you know why two CI are not run and do not have a Details link ?

hungpham2511 commented 4 years ago

I wonder that as well. Probably because I renamed the workflows in the last PRs.

hungpham2511 commented 4 years ago

Also, I think it's worth mentioning that you should use the INTERPOLATION constraint scheme. It provides significantly better numerical stability and solution quality.

jmirabel commented 4 years ago

Same as #115, if you are ok with the changes, I propose to bypass the not-run CI and merge this.

jmirabel commented 4 years ago

I updated this PR as follows:

boundary issues are reported as failure (same behavior as before this PR, with some additional logs),
I added a failing test case.
this PR also contains a few minor unrelated bug fix.

jmirabel commented 4 years ago

I updated and re-based this PR. Is there anything else I should do ?

hungpham2511 commented 4 years ago

I updated and re-based this PR. Is there anything else I should do ?

The PR looks good to me. If you are ok, feel free to merge after CI passes. Do you still have problem with the failing test?

jmirabel commented 4 years ago

The test passes now. Thank you for your help !

jmirabel commented 4 years ago

I may do the same thing in the GLPK solver later.