lang/janet: Update to 1.42.0
ClosedPublic
Actions

Authored by dave_freedave.net on Thu, Sep 3, 3:07 PM.

Details

Reviewers

hsw_bitmark.com
adamw

Commits

R11:5c01d7a330c3: lang/janet: Update to 1.42.0

Summary

I noticed on Janet zulip that they added filewatch via kqueue(2) to janet when creating this version.

I thought (but was wrong) that we had inotify(2) on all supported FreeBSD versions but that wont be true until next year. Anyway you get better file watch support for all the reasons given here.

But that requires patching here instead of upstream. sigh.

Test Plan

Manually run testport with -i and check that for stable/14 it does not have inotify_init1 symbol and that it does have it for stable/15:

poudriere testport -j FreeBSD14 -p local -i lang/janet
nm /wrkdirs/usr/ports/lang/janet/work/janet-1.42.0/_build/libjanet.so | grep inotify_init1

poudriere testport -j FreeBSD14 -p local -i lang/janet
nm /wrkdirs/usr/ports/lang/janet/work/janet-1.42.0/_build/libjanet.so | grep inotify_init1
                 U inotify_init1

logs for stable/14:

fbsd14-janet-1.42.0.log41 KBDownload

logs for stable/15:

fbsd15-janet-1.42.0.log40 KBDownload

And of course make test passes for both.

Diff Detail

Repository

R11 FreeBSD ports repository

Lint

Lint Skipped

Unit

Tests Skipped

Event Timeline

dave_freedave.net requested review of this revision.Thu, Sep 3, 3:07 PM

dave_freedave.net created this revision.

dave_freedave.net added a comment.Thu, Sep 3, 6:19 PM

This comment was removed by dave_freedave.net.

have to use bsd.port.pre.mk and bsd.port.post.mk or it doesn't have INOTIFY_PORT set.

dave_freedave.net attached a referenced file: F170193468: fbsd15-janet-1.42.0.log. (Show Details)Thu, Sep 3, 7:46 PM

dave_freedave.net attached a referenced file: F170193442: fbsd14-janet-1.42.0.log. (Show Details)

dave_freedave.net edited the test plan for this revision. (Show Details)Thu, Sep 3, 7:49 PM

In D59342#1361879, @dave_freedave.net wrote:

have to use bsd.port.pre.mk and bsd.port.post.mk or it doesn't have INOTIFY_PORT set.

Yeah, variables in bsd.port.mk works OOTB because they're present on the initial evaluation pass. The USES macros are included programmatically, so something from b.p.m needs a chance to run first. In general, if a variable comes from a USES macro, you need bsd.port.pre.mk (or, interestingly, bsd.port.options.mk) before reading it.

lang/janet/Makefile
24	Stylistically, when a target exists in both branches of a conditional, we generally prefer to have a single target with the conditional on the inside: post-patch: .if .else .endif rather than: .if post-patch: .else post-patch: .endif
lang/janet/files/patch-src_core_filewatch.c
16	This is a pretty cool approach. Just wanted to get that on the record.

fixed Makefile as suggested to have one target post-patch with .if/.else/.endif in the target body.

I can't believe I spent my life with makefiles that didn't have all this... so I still instinctively do things the hard way 😦

dave_freedave.net marked 2 inline comments as done.Fri, Sep 4, 4:44 PM

dave_freedave.net added inline comments.

lang/janet/files/patch-src_core_filewatch.c
16	This is a pretty cool approach. Just wanted to get that on the record. This is almost identical, just with `%` variables, to what I proposed for upstream. I was just a few months too early but it seems like Janet will take this when 14.4 hits EOL.