Move to tools/tools

Address kib:

• Restart state is one enum (ABI -> SYMLINK or ABI -> NATIVE; never backwards; KASSERT).
• target_resolved is the readable flag; suffix length is aux only. Relative symlink no longer special-cases the retry.
• uma_zfree handles NULL; drop the extra checks.

In D59318#1361056, @fuz wrote:

Good job!

In D60035#1377674, @kib wrote:

Extract chroot_unpriv() from kern_chroot.
Use the same privilege checks for chaltroot(2) as for chroot(2).

Use DISTVERSION (fuz, jrm), add TEST_DEPENDS (jrm)

Rebase

Restore GitHub WWW -- CRAN does not link to GitHub

Use DISTVERSION, remove WWW (use cran.mk default), sort LICENSE,
use BUILD_RUN_DEPENDS

Use DISTVERSION, sort LICENSE, remove WWW (use cran.mk default)

In D59943#1376194, @fuz wrote:

I don't get the whole thing, but it certainly looks reasonable!

Rebase onto origin/main, remove python run dep

Yes, expose it. D59990 can configure the overlay from a native process, dropping the linux_base skip.

Add reproduction step comments to tests

In D59990#1376636, @kib wrote:

I did not read the details of the tests, but I do not see a reason not to commit them now.

Still I have one suggestion (or request if you prefer that name). Could you please annotate each test with expanded file structures that describe the disposition of the directories and symlinks created by tests. I mean, I might want to debug the stuff instead of just running tests, and for that I want to be able to recreate the test configuration of files without running the kyua machinery. Then not having to decipher the innards of the test shell scripts would save energy.

Bump PORTREVISION (fuz), add description for TESTS (portlint)

Bump PORTREVISION as requested. Add TESTS_DESC so portlint is clean.

Tests moved to D59990

tests-only review created as D59990

In D59715#1370517, @fuz wrote:

Consider making installation of the example files conditional on an EXAMPLES option.
LGTM and approved for commit.

In D58183#1375028, @ngie wrote:

There are a lot of lines in this package -- I'm guessing the maintenance is tricky.
Given our tech stack requirements, it might be a good idea to consider lua instead of /bin/sh. Are you open to that possibility (given infinite time/resources of course..)?

Looks great! Approved. Give fuz a day or two to chime in.

Address review comments from fuz:

Rebase onto head

@fuz I created D59867 for follow-up to implement optional items mentioned

testport ... success
portclippy ... happy
portfmt -D ... ok
portlint ... fine

Fix distinfo following upstream asset swap

poudriere testport caught upstream shenanigans. Brave re-uploaded the v1.94.117 release assets after f4f2f98ba7d8 bump was committed by fluffy

testport ... failed (working on fix for failed checksum step)
portclippy ... happy
portfmt -D ... ok
portlint ... fine

• Rebase atop f4f2f98ba7d8dc542a967e3ba7d7f1231624792c (fluffy)
• Utilize ARCH:S for EXTRACT_SUFX (fuz)
• Replace PORT_OPTIONS:MWIDEVINE with option-target (fuz)
• Satisfy portlint, clamp pkg-descr to 80c, move DISTFILES

testport ... passed
portclippy ... happy
portfmt -D ... ok
portlint ... fine

Update to 3.2

I am decoupling this from D58855 because the issue as far as I can see is that without D58855, only 16.0-CURRENT users will be unable to run the browser; secondly, D58855 affects a LOT more than just www/linux-brave (without D58855 any linux program that lives outside of emul_path -- /compat/linux -- presently has issues with broken symlinks on 16.0-CURRENT)

I'm thinking of moving these to tools/ or tools/tools/ as I'm not sure it brings much value to have them available in usr.bin/ -- absent any feedback on that, I'll just forge ahead to usr.bin/ as planned. Just spit-balling an idea that popped into my head.

Small fix for object detection on builds without -j

In D59746#1371500, @ziaee wrote:

Since it doesn't do anything, it is not necessary to remove them. However, since it doesn't do anything, it is kinda nice to remove them. To clarify, I was not asking you to remove them, I was just asking where you got the idea to do this because I have never seen anyone do this. Usually weird undocumented stuff comes from copying other files, so that's what I was wondering.

Also approve. Looks great. 👍

In D59714#1371134, @fuz wrote:

In D59714#1371051, @dteske wrote:

In D59714#1370491, @fuz wrote:

Oh I thought we would be removing WWW now that it is the default set by USES=cran.

On the other review you said "Could you do a sed through the ports tree to edit all affected ports automatically?"

So I did a sed to edit all the affected ports.

If you want deletion instead, say the word.

Well with “edit” I meant to edit such that they use the new default WWW from USES=cran, i.e. to delete the explicit assignments. I am sorry that I was unclear here.
Nevertheless I believe your work is quite a good foundation, as we can only remove explicit WWW assignments if they are known to match the default exactly.
We can also check in this patch as is and then delete the WWW assignments in a secondary step, so it can easily be undone if there are complaints.