For reasons @arrowd pointed out changing LOCALBASE back to /usr/local. I incorrectly trusted portlint on those.

remove bad fix, its addressed in D55487.

As @des has the correct fix for ngctl, I'm going to just going to update this to have my manpage bug. Thanks!

The usage is, I believe, correct:

# ngctl -h
ngctl: illegal option -- h
usage: ngctl [-j jail] [-d] [-f filename] [-n nodename] [command [argument ...]]

add NO_ARCH=yes

Just ran poudriere testport -j FreeBSD15 -p local lang/jpm and could probably use NO_ARCH=yes like D54667.

Thanks Adam! I think its ready for commit now.

fix style

Created PR for bee.lua. But even once fixed, there isn't any need for the tests to run during port build and are still available (as I still want to run them on updates) via make test.

Once I realized the test failure isn't coming from do-test I realized I didn't need a maintainer-test target.

Moved LM_INOTIFY_LINK= above all the rules.
No longer run luamake tests as part of build. Only run for explicitly requested make test as they have a flaky (actually broken) test too.

Actually I just hit the same test failure pkg-fallout had emailed me about. bee has a test here that I don't believe is correct.

All suggestions taken:

• always install pkg-message
• use bsd.port.{pre,post}.mk as suggested
• update pkg-message to explain undefined symbols.

Latest output from poudriere testport -j FreeBSD15 -p local devel/lua-language-server
{F144553961}

Moved to using the new USES=inotify, verified it worked:

stable/14 » pkg info -d lua-language-server
lua-language-server-3.17.1,1:
        libinotify-20240724_3
        libinotify-20240724_3 (libinotify.so.0)

update synopsis as well. Here is the new output for /bin/sh /usr/share/examples/bhyve/vmrun.sh -h:

Usage: vmrun.sh [-aAEhiTuvw] [-9 <9p share>=<path>[,<opts>]]
                [-c <CPUs>] [-C <console>] [-d <disk file>]
                [-e <name=value>] [-f <path of firmware>] [-F <size>]
                [-G [w][address:]port] [-H <directory>]
                [-I <location of installation iso>] [-l <loader>]
                [-L <VNC IP for UEFI framebuffer>]
                [-m <memsize>] [-n <network adapter emulation type>]
                [-p <pcidev|bus/slot/func>]
                [-P <port>] [-t <tapdev>] <vmname>

fix unititialized i=... (I had that on my local copy and somehow lost it).
add usage for -9. I'm open to better suggestions, though.

adding 9: to opts for arm64 and re-creating diff with right options.

re-ran poudriere testport -j FreeBSD15 -p local devel/lua-language-server

lua-language-server has updated before this could be committed. skipping ahead to latest release 3.17.1

I decided to change a variable name to match the Uses/inotify.mk coming in D54116. That really just meant changing LINK_INOTIFY to INOTIFY_LIB. But while at it I wanted a better variable name for what I had set as INOTIFY, so it is now LM_INOTIFY_LINK.

output from poudriere testport ... devel/janet-lsp: {F141925749}

{F141925403} is output from poudriere testport ... devel/lua-language-server.

Confirmed via poudriere testport that USES does not need lua:build as it only uses its own modified bee.lua which is now based on lua 5.5.

Removed files from patch that didn't actually change (just date change from when I ran make makepatch).

The jimage file is truly architecture independent. Not just amd64 vs arm64. But it works for FreeBSD, Linux, Windows, basically anything that has a janet interpreter.
The jdn file is the janet equivalent of edn which is the clojure equivalent of JSON. It is just recording what is inside the jimage.

Thanks for adding RUN_DEPENDS to port.

I had some bad grammer in janet-lsp.in comments. This just fixed that.

Address review comments:

• place .for at line beginning.
• use LOCALBASE to invoke jpm.
• update pkg-descr

{F139748281}
poudriere-testport(8) output, were it not for this I would have missed the git rev-parse in the janet-lsp code!

switched to using ports-mgmt/submodules2tuple to do the initial pass.

Yeah using the submodule2tuple shows I neglected a submodule:
actboy168:ltest:0d2d81f:ltest/3rd/json.lua/test/ltest

Actually I do have portlint -A outpout:

WARN: Makefile: new ports should not set PORTEPOCH.

{F139313289} log file from poudriere testport -j 15R -p local devel/lua-language-server

In D54055#1236567, @adamw wrote:

There it is! It's looking great, Dave. I'm just doing some final tests and--barring anything wacky--I'll get it committed shortly.

Those empty directories are LuaCATS modules. They are extra type annotations for popular libraries. They don't get built (or tested it would seem) but they are nice to have.

Looking into it. I just got poudriere running in a bhyve VM. I can see the same errors so next update should be clean.

changed to check for inotify the way dns/aardvark-dns does. it passes both portclippy and portlint and keeps all the variables in one place.

one sec I'm changing it to the way requested because both portlint and portclippy like it. so it has that plus grep going for it.

If only I'd been slower. I prefer the file check but can add comment. Before I do though, is the check in 2 places acceptable?

I didn't get nearly as many errors from portlint. But it may be I fixed the end of line spaces first. Anyway both portclippy --strict and portlint -A seem happy now.

I only ran portclippy before. I'll go find portlint and track down the other errors too.

In D54055#1236171, @diizzy wrote:

Why not use https://github.com/LuaLS/lua-language-server/releases/download/3.16.0/lua-language-server-3.16.0-submodules.zip ?

Change to mimic strategy used in devel/glib20. Also I think portclippy didn't see the variables on stable/15 before so I had to change them to have leading '_' and at that point they can be up by LIB_DEPENDS.

Oh relying upon the header file would also help in case inotify gets backported to stable/14 in the future.

removed change to ports/devel/Makefile

Not that I doubted you, but I did rebuild after removing GH_PROJECT and GH_TAGNAME.

adopt review comments

Most up to date version is at https://github.com/dmarker/bong-kmods

Most up to date version is at https://github.com/dmarker/bong-utils

While I will, of course, be very happy if this makes it into FreeBSD, I am more happy that D50241 has made it in.
This is because I can easily maintain this outside the src tree and have prepared for that eventuality.

In D50241#1194014, @zlei wrote:

@dave_freedave.net Thanks for your contribution !

Is there anything more that needs to be addressed here to add a -j option to ngctl(8) ?

In D50244#1156230, @bz wrote:

I do understand the problem. I just think the name is a bit futuristic and not very descriptive for what it is doing.

I'm calling this "done" but I'm still open to a name change.

remove NG_WORMHOLE_WARP from ng_wormhole.h as its not needed and in fact can't be referred to this way.

Marking changes addressed done. Just realized I still need to get "warp" and its define moved to C file.

Just using VIMAGE in the Makefile didn't work but I noticed sys/modules/Makefile had an example that does work for wtap. I also stole the assert from sys/dev/wtap/if_wtap.c.

I didn't think the empty string would be passed along, but @kevans had it right. I moved the check to later before switching to jail.

Remove check for NULL from -j option not required. Check for empty string before attempting to switch to jail.
New output:

root@fbsd15:~ # ngctl -j
ngctl: option requires an argument -- j
usage: ngctl [-j jail] [-d] [-f filename] [-n nodename] [command [argument ...]]
root@fbsd15:~ # ngctl -j ''
ngctl: invalid jail name
usage: ngctl [-j jail] [-d] [-f filename] [-n nodename] [command [argument ...]]

Document that you must leave the jail empty to indicate the jail where you run ngportal.

Allow a jail to be specified numerically to the "open" command.

Be consistent with other commands and have [-j jail] first. Clean up other issues in man page. Importantly the usage string and the man page synopsis agree now.

In D50244#1153881, @imp wrote:

What is the use case for this?

Fixed man page to be mandoc -Tlint compliant.
Saw a style issue in Usage() and fixed it.
parse_spec() was giving errors about "component[x]" which has no meaning at all to users of ngportal and was changed to display the component with an error which should be much more helpful.

I saw that other man pages used the simplified license so I changed this to do the same.