Project

General

Profile

Actions

Pull request #429

closed

Gepetto updates

Added by Guilhem Saurel about 2 months ago. Updated about 1 month ago.

Status:
Closed
Priority:
Normal
Assignee:
-
Repository URL:
https://github.com/nim65s/robotpkg
Repository branch:
master

Description

Hello !

Here are updates for some gepetto packages:
- eiquadprog 1.2.9
- eigenpy 3.10.0
- hpp-fcl 2.4.5
- pinocchio 3.2.0
- TSID 1.7.1
- HPP 5.2.0

It also add some update for:
- sysdep/numpy v2
- urdfdom-headers 1.1.2
- console-bridge 1.0.2
- urdfdom 4.0.1
- casadi 3.6.7 (and split the package in casadi & py-casadi, and also activate casadi support in pinocchio)

I did test this on Arch, 20.04 & 22.04

Cheers,
Guilhem

Actions #1

Updated by Anthony Mallet about 2 months ago

  • Status changed from New to Feedback

Thanks, pushed

Actions #2

Updated by Guilhem Saurel about 1 month ago

Hello,

Andrea reported a failure on py-tsid apt package in https://github.com/stack-of-tasks/tsid/issues/243

If fixed it in https://github.com/stack-of-tasks/tsid/issues/243, and pushed that as a patch on my robotpkg repository, but I don't have a working robotpkg bootstrap available, so the updated checksum is missing.

I also did not bump PKGREVISION, because the package is failing on hydra anyways, but I'm not sure about that.

Looking at hydra results, other issues from this PR include:
- a PLIST warning on pinocchio. I guess we can ignore that until the next release, where I will try to remember to fix this.
- a PLIST warning on eigenpy, only on NetBSD. I'm not sure where this .orig is coming from, but I guess this is not very high priority.
- urdfdom build error on Debian 11. Maybe an issue with tinyxml version, I'll try to have a look, but next week.
- crocoddyl build error on NetBSD. seems related to ipopt. Easiest workaround would be to deactivate ipopt feature on that platform I guess :)

Best,
Guilhem.

Actions #3

Updated by Anthony Mallet about 1 month ago

On Monday 28 Oct 2024, at 19:54, Guilhem Saurel wrote:

Andrea reported a failure on py-tsid apt package in
https://github.com/stack-of-tasks/tsid/issues/243

Wrong PLIST (missing files) it not a fatal error when you do `make
install` (or `update` for that matter) in robotpkg, as this would
prevent to easily (re)generate the PLIST. But it is a fatal error when
you do 'make package`, so maybe this can become your favorite target
if/when you are doing tests :)

I tried to have a look myself when I saw that error, but I got lost in
the myriads of cmake files :)

If fixed it in https://github.com/stack-of-tasks/tsid/issues/243,
and pushed that as a patch on my robotpkg repository, but I don't
have a working robotpkg bootstrap available, so the updated checksum
is missing.

OK, fixed.

I also did not bump PKGREVISION, because the package is failing on
hydra anyways, but I'm not sure about that.

As you said it's always been broken, so keeping PKGREVISION unchanged
will not hurt. Though it nevers hurts to bump it either!

Looking at hydra results, other issues from this PR include:
- a PLIST warning on pinocchio. I guess we can ignore that until the
next release,

Not sure if that pinocchiopy.pc should be there or not, but adding it
to the PLIST is trivial and will prevent it to be leftover if people
install from source.

- a PLIST warning on eigenpy, only on NetBSD. I'm not sure where
this .orig is
coming from, but I guess this is not very high priority.

the `patch` command on NetBSD creates a backup file by default, so this
.orig comes from the fact that FindBoost is patched. Then the cmake
file installs whatever it finds in the cmake dir, leaving this
unwanted file behind.

- urdfdom build error on Debian 11. Maybe an issue with tinyxml
version, I'll try to
have a look, but next week.
- crocoddyl build error on NetBSD. seems related to ipopt. Easiest
workaround would
be to deactivate ipopt feature on that platform I guess :)

Strange, that may indeed be an issue with ipopt not linked properly
with lapack. I can have a look, but no big deal indeed.

Actions #4

Updated by Guilhem Saurel about 1 month ago

  • Status changed from Feedback to Closed

Thanks a lot, http://robotpkg.openrobots.org/rbulk/robotpkg/broken.html is in a way better shape now !

I'll switch my test process to `make package`, thanks for the hint :)

I think this pinocchiopy.pc is a mistake. Using this prevent us from having multiple python bindings installed. it could be eg. pinocchiopy312.pc, but I'm not sure if this is a good idea. Actually I don't really see the point for this file overall.

Actions

Also available in: Atom PDF