With a bit of luck it'll work reasonably well.

In D41610#948388, @tcberner wrote:

Moin moin

portmgr values the idea of upstream CI. But given the recent branching of 14 from CURRENT, this looks like a safe enough change, to go forward under the "just fix it " blanket.

mfg Tobias/portmgr

In D41610#948383, @des wrote:

In D41610#948373, @pjfloyd_wanadoo.fr wrote:

You presume that it will be no worse WHEN NOBODY HAS EVEN RUN THE VALGRIND TEST SUITE.

FreeBSD 14 and 15 are essentially identical. We just bumped the version number last week. The test suite produces the exact same result on both:

== 813 tests, 477 stderr failures, 83 stdout failures, 11 stderrB failures, 17 stdoutB failures, 37 post failures ==

I suggest you reevaluate your attitude.

In D41610#948371, @des wrote:

In D41610#948370, @pjfloyd_wanadoo.fr wrote:

Right now if you really need to use Valgrind on FreeBSD 15 [...]

My goal is not to use Valgrind, my goal is to fix a broken port. At the moment, this patch is all that is needed to make devel/valgrind build and run no better or worse on FreeBSD 15 than on FreeBSD 14.

In D41610#948369, @des wrote:

In D41610#948362, @pjfloyd_wanadoo.fr wrote:

When I have VMs to test on FreeBSD 15

You can easily upgrade a FreeBSD 14 VM to FreeBSD 15.

Right now if you really need to use Valgrind on FreeBSD 15 the best thing to do would be

In D41610#948361, @des wrote:

In D41610#948348, @pjfloyd_wanadoo.fr wrote:

In D41610#948312, @des wrote:

Right now FreeBSD 15 is essentially identical to FreeBSD 14 except for the version number.

Really?

Yes, really.

In D41610#948355, @dim wrote:

Note, this would also apply to devel/valgrind-devel.

In D41610#948344, @emaste wrote:

FreeBSD 15.0 isn't even available for download yet.

Snapshots will start being available soon, and a decent fraction of the FreeBSD developer community is already running 15 or will be shortly.

In D41610#948312, @des wrote:

FreeBSD 15.0 isn't even available for download yet.

Whilst that should work my plans were to do the following

LGTM

In D37201#844760, @jrm wrote:

To be clear, I'm just referring to the style changes mostly to conform to Chapter 15. Order of Variables in Port Makefiles and a few other style issues. It's kind of like style(9) for ports. No problem if you prefer to leave things as is, but could you clarify?

In D37201#844432, @jrm wrote:

Also, if you are good with these changes, I can do the same for devel/valgrind.

Yes, please add the devel/valgrind-devel patch.

I've pushed a couple of changes upstream. Could you try building from source

I have a commit bit for the Valgrind source repo and I prefer to make changes directly there rather than maintain patches in the ports tree.

I've put much more detailed regtest failure analysis here

Accepted

Looks OK now. make && make install worked, running the regression suite on the installed Valgrind gave me expected redults.

I had a go at building, replacing valgrind-dev Makefile and distinfo. I got