Page MenuHomeFreeBSD

Cleanup ports after 11.3-EOL
ClosedPublic

Authored by rene on Oct 2 2020, 8:58 PM.
Tags
None
Referenced Files
Unknown Object (File)
Tue, Dec 17, 1:08 PM
Unknown Object (File)
Mon, Dec 9, 4:45 AM
Unknown Object (File)
Nov 8 2024, 5:14 PM
Unknown Object (File)
Oct 26 2024, 5:42 PM
Unknown Object (File)
Oct 26 2024, 5:41 PM
Unknown Object (File)
Oct 26 2024, 5:40 PM
Unknown Object (File)
Oct 26 2024, 5:40 PM
Unknown Object (File)
Oct 26 2024, 5:39 PM

Details

Summary

Mk/bsd.port.mk: bump minimum supported version of FreeBSD to 11.4

Bump versions for ports referencing 11.3 using "git grep 1103"

Remove some 12.0 compatibility too while here.

www/chromium: Remove now obsolete instructions for hanging tabs

r337328 was in 12-CURRENT and has been MFC'd to 11.2-STABLE

sysutils/p5-BSD-Jail-Object: no need to differentiate between FreeBSD 7 and 8 anymore.

emulators/mesen: add some patches originally proposed for r516983

Merge branch 'main' into 11.3-cleanup

Merge branch 'main' into 11.3-cleanup

Merge branch 'main' into 11.3-cleanup

more 11.3 cleanup

#git grep 11\\.3 | grep -vw distinfo | grep -vw pkg-plist | manualfilter

Diff Detail

Lint
No Lint Coverage
Unit
No Test Coverage
Build Status
Buildable 34312
Build 31445: arc lint + arc unit

Event Timeline

rene requested review of this revision.Oct 2 2020, 8:58 PM

The diff to Mk/bsd.port.mk might be a bit stale, I based this patch on the cgit-beta main branch.

rene retitled this revision from Mk/bsd.port.mk: bump minimum supported version of FreeBSD to 11.4 to Cleanup ports after 11.3-EOL.Oct 2 2020, 9:50 PM
rene edited the summary of this revision. (Show Details)
rene edited reviewers, added: x11; removed: zeising.
multimedia/libva/Makefile
29 ↗(On Diff #77802)

Can you drop above 2 lines as well? Nothing needs flags specific to OS version along with the comment documenting why it was there.

Rene, for graphics/rawtherapee I will need to investigate first. Thanks for the reminder.

graphics/rawtherapee/Makefile
101 ↗(On Diff #77802)

Please do not update this line, AFAIR 12.0 was fine and there is no need to update this check and skew history.

zeising added inline comments.
graphics/drm-kmod/Makefile
22

This comment is technically not true.
I'm not sure if it'suseful to update the version check to say 12.1, or if it can just be left as is.

mandree requested changes to this revision.Oct 3 2020, 10:06 AM

Rationale first, then the request in the bottom three lines in boldface.

(11:59:16) mandree: updating version ranges in Makefiles does NOT make much sense to me. Removing stuff marked "remove after 11.3 EOL" DOES make sense, however.
(11:59:41) rene0: this can be a relatively sweeping patch by definition, because the removed code is never activated again
(12:00:37) mandree: but why revise historic information? That's useless. The feature and component versions of the historic now-end-of-life do not change.
(12:00:47) mandree: rene0: no, it cannot
(12:01:01) mandree: it forges history, that's why I am objecting quite strongly, as seen above.
(12:01:40) rene0: the history is still in svn/git
(12:01:49) mandree: No need to forge Makefiles
(12:02:02) rene0: the port itself is a recipe, not an official record of events
(12:02:08) mandree: you essentially make Makefile and SVN history fall apart, and that's frowned upon
(12:03:47) rene0: the ports can still be comitted individually
(12:05:47) mandree: so let me put it actionable: the rawtherapee and ebsnvme-id and linux-c7-libdrm and drm-kmod and i386-wine patches are all wrong. Feel free to update the IGNORE lines but not the .if lines.
(12:05:51) mandree: The rust-bootstrap and gnatcross-sysroot-aarch64 patch may be dangerous because you bring new assumptions about what the actual port does with the release information.
(12:05:51) mandree: Make run-time tests before/after the patch, consult the experts, and DO NOT COMMIT unless you have EXPLICIT approval.

This revision now requires changes to proceed.Oct 3 2020, 10:06 AM

I am adding a few inline comments (not thorough), please review them and then omit blunt sweeping OSVERSION changes.

Mk/bsd.port.mk
1180 ↗(On Diff #77802)

This is the only place where OSVERSION checks need to be updated.

emulators/i386-wine-devel/Makefile.amd64
35

This update on the .if line should NOT be made. The one on the following IGNORE line is fine though.

emulators/i386-wine/Makefile.amd64
36

This update on the .if line should NOT be made. Changing IGNORE is fine.

graphics/drm-kmod/Makefile
19

Please do not update this - versions >= 1200080 and < 1201000 do not magically change their behaviour, so again this check should not be changed. Even if you think that it can no longer be activated, let's not make review of "why did we did this or that in the past" kind more difficult and mislead people reading history.

I'll need to test the new generated rust bootstrap, when do you plan to land it?

I'll need to test the new generated rust bootstrap, when do you plan to land it?

Well, given all the you're-doing-it-wrong hostility on #bsdports instead of just explaining things, maybe never.

This revision was not accepted when it landed; it landed in state Needs Revision.Oct 5 2020, 7:52 PM
This revision was automatically updated to reflect the committed changes.

uh, this should not have been closed - the commit tracking interface is a bit too simplistic and is not up for partial commits. Reopening.

uh, this should not have been closed

Yes do not list the review in a commit Differential Revision unless it should be closed. You can list the D# in the commit message and Phab will add a link between the commit and the review - e.g. something like "This commit is part of review D26646" or "Reported by: rene in review D26646"

@rene will you re-upload the diff to put it back in a known stte

uh, this should not have been closed

Yes do not list the review in a commit Differential Revision unless it should be closed. You can list the D# in the commit message and Phab will add a link between the commit and the review - e.g. something like "This commit is part of review D26646" or "Reported by: rene in review D26646"

@rene will you re-upload the diff to put it back in a known stte

Hmmm, I've always just listed it multiple commits and reopend and/or not cared about the state of the review (though usually when I do that, it's on boring reviews that get few if any comments, not something as contentions as this one)...

Rebase on ee5204859b9328a9f189d2912812add11ccf73af

  • Mk/bsd.port.mk: bump minimum supported version of FreeBSD to 11.4
  • Bump versions for ports referencing 11.3 using "git grep 1103"
  • www/chromium: Remove now obsolete instructions for hanging tabs
  • sysutils/p5-BSD-Jail-Object: no need to differentiate between FreeBSD 7 and 8 anymore.
  • emulators/mesen: add some patches originally proposed for r516983
  • Merge branch 'main' into 11.3-cleanup
  • Merge branch 'main' into 11.3-cleanup
  • Merge branch 'main' into 11.3-cleanup
  • more 11.3 cleanup
  • Merge branch 'main' into 11.3-cleanup

Ah, arc diff likes to repeat its previous "commit" log.

The hash is that of the main branch of https://cgit-beta.freebsd.org/ports.git

I'll need to test the new generated rust bootstrap, when do you plan to land it?

Did you manage to? 1.47 update was a good opportunity.

  • Merge branch 'main' into 11.3-cleanup, rebase upon 98a89e70ef40

I'll need to test the new generated rust bootstrap, when do you plan to land it?

Did you manage to? 1.47 update was a good opportunity.

I left rust at 11.3 for now.

@rene, can you address libva inline comment?

rene marked an inline comment as done.
  • multimedia/libva: drop two additional obsolete lines
rene marked an inline comment as done.Oct 17 2020, 3:00 PM

Applied the patches which should be politically safe.

I'll need to test the new generated rust bootstrap, when do you plan to land it?

Did you manage to? 1.47 update was a good opportunity.

No, too many fallout with the1.47 update, I couldn't find the time to try to update the bootstrap.

mandree requested changes to this revision.Oct 17 2020, 4:02 PM

@rene can you PLEASE stop editing all the many .if lines and updating them from >= [11.3] to >= [11.4] and similar, in particular (1) graphics/drm-kmod/ (2) graphics/linux-c7-libdrm and (3) sysutils/ebsnvme-id
I repeat, these are .if lines that check behaviour for past versions of FreeBSD - and their past behaviour does not change with 11.3 becoming EOL, it does not make sense to update these version limits, it will only be confusing for people when they debug and touch history.

This revision now requires changes to proceed.Oct 17 2020, 4:02 PM

Please also see my new inline comments above.

emulators/i386-wine-devel/Makefile.amd64
35

This update on the .if line should NOT be made. The one on the following IGNORE line is fine though.

I withdraw that, I was misreading the intentions when commenting for the first time.

emulators/i386-wine/Makefile.amd64
36

This update on the .if line should NOT be made. Changing IGNORE is fine.

^ please disregard (see i386-wine-devel, same reason)

graphics/drm-kmod/Makefile
19

Please do not update this - versions >= 1200080 and < 1201000 do not magically change their behaviour, so again this check should not be changed. Even if you think that it can no longer be activated, let's not make review of "why did we did this or that in the past" kind more difficult and mislead people reading history.

This comment stands, please drop this change. The behaviour of, say, exactly 1200080, does not change, so we should not clobber this knowledge of when *exactly* the kmod became needed. It will only confuse people trying to understand "why 1201000" because the OSVERSION you check is now detached from the actual OSVERSION that correlates with the change.

graphics/linux-c7-libdrm/Makefile
22

Please omit this change. We gain exactly nothing, except confusion because we lose the connection between the OSVERSION that brought the relevant change and the new .if check.

sysutils/ebsnvme-id/Makefile
25

This change does not make sense. Either we leave it as is to document the behavior of historic FreeBSD versions, or we change it all the way and also remove the OSVERSION < 1104000 part, since Mk/bsd.port.mk should handle it already through the UNSUPPORTED... checks.

rene marked 5 inline comments as done.
  • Merge branch 'main' into 11.3-cleanup
  • Merge branch 'main' into 11.3-cleanup
  • graphics/dspdfviewer: simplify logic for -Wno-error=extra-semi-stmt
  • simplify check for supported versions after 11.3 EOL
graphics/drm-kmod/Makefile
19

@rene Again, please omit these changes to graphics/drm-kmod/Makefile, as @zeising also pointed out. You are checking features of historic kernels here, not ports tree, and existing kernel features don't change because someone pulls the plug on the ports tree. You don't really attain simplification here, just leave this file untouched. Let's clean this up after the entire 11 branch is EOL.

mandree requested changes to this revision.Oct 22 2020, 7:13 PM
This revision now requires changes to proceed.Oct 22 2020, 7:13 PM
This revision was not accepted when it landed; it landed in state Needs Revision.Oct 28 2020, 10:49 AM
This revision was automatically updated to reflect the committed changes.