Community Packages

Please read this before reporting a bug:
https://wiki.archlinux.org/title/Bug_reporting_guidelines

Do NOT report bugs when a package is just outdated, or it is in the AUR. Use the 'flag out of date' link on the package page, or the Mailing List.

REPEAT: Do NOT report bugs for outdated packages!
Tasklist

FS#68585 - [opencascade] please add BUILD_RELEASE_DISABLE_EXCEPTIONS=OFF

Attached to Project: Community Packages
Opened by M. Greyson Christoforo (greyltc) - Tuesday, 10 November 2020, 17:18 GMT
Last edited by Kyle Keen (keenerd) - Tuesday, 24 November 2020, 00:08 GMT
Task Type Feature Request
Category Packages
Status Closed
Assigned To Kyle Keen (keenerd)
Architecture All
Severity Low
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 0
Private No

Details

I'd like
BUILD_RELEASE_DISABLE_EXCEPTIONS=OFF
here please.

It seems like this bit of software I'm trying to package for the AUR needs that build flag:
https://github.com/tpaviot/pythonocc-core/issues/916
This task depends upon

Closed by  Kyle Keen (keenerd)
Tuesday, 24 November 2020, 00:08 GMT
Reason for closing:  Fixed
Additional comments about closing:  opencascade-7.5.0-2
Comment by Kyle Keen (keenerd) - Thursday, 19 November 2020, 03:49 GMT
Hold up, for a minute. What exactly are you trying to package for the AUR? python-occ-core is already officially packaged: https://www.archlinux.org/packages/community/x86_64/python-occ-core/

----------

So, the actual bug report:

This was previously the default for Opencascade, until March 2019. Then it was disabled in Release builds for performance reasons, while still being enabled for Debug builds.

There are four options:

* Change Arch's package.
* Build your own opencascade-debug package.
* Fix pythonocc to support the principle library it uses.
* Convince Opencascade to roll back this performance change.

Personally, I think that changing Arch's package is a little silly. You'll run into the same issue across every linux distribution, probably.

However, this is complicated by python-occ-core being an officially packaged package. If we'll run into the problem regardless, then we'll need to make it work regardless. Python-occ-core is FFY00's package, let me see what he has to say.
Comment by Filipe LaĆ­ns (FFY00) - Sunday, 22 November 2020, 19:33 GMT
I would prefer this change to be applied. Unfortunately, I don't think we have much of an alternative, at least something reasonable. If there is a significant performance hit, we could always have special opencascade build, but I do not think that's the case.
Comment by Kyle Keen (keenerd) - Monday, 23 November 2020, 03:47 GMT
How is opencascade-7.5.0-2 for you?
Comment by M. Greyson Christoforo (greyltc) - Monday, 23 November 2020, 11:30 GMT
Oh no! Sorry, I didn't realize my AUR package, https://aur.archlinux.org/packages/python-occ/ was a duplicate of an official one!
To be fair though, maybe the official one was a dupe of mine. I first put mine in the AUR in 2016. It's probably my fault though for manipulating its name.
No matter, I've filed a request for its deletion :-)

I'll address opencascade-7.5 in general (off topic here): the recent bump from 7.4 to 7.5 broke a few things that depend on it. Most notably freecad. Freecad now needs opencascade74 which conflicts with opencascade so it's no longer possible to have the official kicad and freecad packages on the same machine. That's bad because it breaks a bunch of typical hardware dev workflows (like mine). My workaround is to install freecad-git from the AUR instead of freecad from community. Also, that version bump doesn't break kicad, but it does mean kicad needs to be rebuilt to work with opencascade-7.5.

Now to finally actually address the recent change in opencascade-7.5.0-2 that this bug report is about, thank you, the new build flag looks like it fixed the issue and so I'll close the bug I filed with upstream.

On a side note @FFY00, you might consider adding a check() to your python-occ-core PKGBUILD to run upstream's test suite. I only noticed this problem because of the check() I had in mine. I had to do a silly hack to get it to work, I'm happy to help!
Comment by Kyle Keen (keenerd) - Tuesday, 24 November 2020, 00:07 GMT
Glad it works.

As a postscript, I wouldn't close the upstream bug. A piece of software should really support the default configuration of the libraries it uses. Building with this flag is a workaround, one that would need to be done everywhere.

Loading...