2026-01-21T15:15:37-0800 Andrew Hewus Fresh <[email protected]>:
> On Sun, Jan 18, 2026 at 12:47:52PM +0000, Johannes Thyssen Tishman wrote:
> > 2026-01-17T18:26:12-0600 [email protected]:
> > > Update cad/opencascade from 7.8.1 -> 7.9.3
> > > 
> > > Tested with kicad 9.0.6 on amd64
> > 
> > Hi James,
> > 
> > Thanks for the update. I've actually had a very similar diff to yours
> > for a while now (with a few additions to allow running tests), but I
> > just couldn't find the time/motivation to finish it. I find this port to
> > be quite messy and complicated, and every time I work on it I really
> > need to spend a few hours trying to understand it. I'm sorry if this
> > caused duplicate work.
> > 
> > Anyways, below is a slightly modified version of your diff, with a
> > couple additions to allow running tests. Tests are still running as I'm
> > writting this email (there are 21082 tests :/), but I'll report back
> > once they are done.
> > 
> > On another note, we (or I as maintainer) still need to test the other
> > consumers of the port:
> > 
> > $ show-reverse-deps cad/opencascade
> > cad/freecad
> > cad/horizon-eda
> > cad/kicad
> > cad/netgen-mesher
> > cad/prusaslicer
> 
> I have _lightly_ tested freecad and prusaslicer with the previous diff
> and haven't had any issues yet.

Thanks for testing afresh1@.

I have built all remaining consumers now without any issues, but I
didn't test runtime. However, I did manage to find the time to run the
tests and got the following results:

1. Test run as _pbuild user with PORTS_PRIVSEP=yes (with xhost
+si:localuser:_pbuild; make test; xhost -si:localuser:_pbuild):

  Total cases: 137 BAD, 1029 FAILED, 1 IMPROVEMENT, 4698 OK, 15217 SKIPPED
  Elapsed time: 2 Hours 38 Minutes 22.977169 Seconds

2. Test run as my own user by running the commands under the 'do-test'
target manually:

  Total cases: 142 BAD, 265 FAILED, 1 IMPROVEMENT, 5457 OK, 15217 SKIPPED
  Elapsed time: 2 Hours 51 Minutes 17.465214 Seconds

Running the tests as the _pbuild caused a lot more failures mostly due
to the following error:

  MESA-LOADER: failed to retrieve device information
  amdgpu: drmGetDevice2 failed.

Not sure how to deal with this. When running the tests as my own user,
ca. 95% of the tests succeed (without counting bad and skipped tests)
and failures are mostly due missing font dependendencies. I'll keep
working on these, but I consider the results to be good enough and would
like to commit this update that is long overdue.

Feedback? Oks? Objections?

On another note, OpenCascade 8.0.0 is just around the corner (currently
at RC3). From what I read, the update comes with a lot of changes to the
codebase and build system, so we'll be back at this soon.

Reply via email to