don't skip testing ports with a suffix (like www/rubygem-rack16)

o fix case with no runtime dependencies (devel/rubygem-extlib)
o speed up by deduplicating a list of version strings to check

Great, thank you! As long as the style is still OK...

handle situation where multiple local versions of the same gem are installed

Thank you!

There *should* be, but that's not the case everywhere in the tree.
For example, www/rubygem-merb-core bundles a bunch of crap, including an older version of itself.

address mat's feedback:

I think I'm OK with this. The upstream chooses to host release tarballs on GitHub and if the maintainer doesn't like the added complexity of fetching multiple things and converting markdown files. This is not a commit approval until matthew chimes in.

I submitted an exp-run request: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=220605

Approved!

Is this close enough to ask for an exp-run at this point?

address mat's feedback:
make ${msg} local

OK, GH_TUPLE sucks... This is fine to go in as-is.

Looks great!

You can mess with these by setting GO_PKGNAME and GO_TARGET.

address mat's feedback

Handbook trumps portlint.

Looks good for now as-is, but just a note that for new ports that don't have actual releases, we prefer versions with a "g" in them, like so: "g20170630".

Looks good!

Convert this and possibly other stuff to USES=go:

You should replace all of this with USES=go stuff:

address mat's feedback:
o feed while loops with heredocs
o use ${msg} instead of $msg

o address mat's feedback
o use pipe continuation style consistent with the rest of the file

The != assignments are like a performance hand-grenade that splatters everyone around it.
The right-hand side expression would get evaluated every time the file is processed....
In this case I see that it's inside of a non-default option and it would be ugly to work around it,
so it might not be so bad. I'll let other people chime in.

As portlint is complaining, move LIB_DEPENDS and TEST_DEPENDS up

Approved!

Looks good!

Approved!

OK, awesome!

make the style more consistent and don't reuse $ret

more descriptive message

roll in qa-gem.sh into qa.sh so that Poudriere can pick it up

update to 0.9.11

OK, thanks for looking into it. This looks good now that the .sample stuff is resolved.

Approved!

Approved!

Is there a reason for removing BROKEN_* stuff?
Did something change that would make the port work
on these platforms or did the submitter test it, etc.?

You need to bump the port revision for dependency changes.

keep maintainer

keep maintainer

Well, my idea was to check for presence.

start_precmd()
{
  if [ -f ${dccd_home}/<my_file> -a -f ${dccd_home}/<other_file> .... ]
  then
    :
  else
    <do stuff>
  fi
  <the rest of start_precmd>
}

You need either a "Reported by:" or "Submitted by:". Otherwise, it's good.

abstract PORTNAME in plist

add LICENSE_FILE

The whole POST-INSTALL target could be rolled into the RC script.

Switch these to /d per mat's feedback and commit vuxml piece first and then immediately the port.
Other than that, looks good!

Looks great!

Looks good!

Approved!

Looks good!

Approved!

Looks good to me!

OK, approved for commit.

Did you verify that it's the same maintainer? If so, approved!

Looks good!

OK, It looks good to me as-is. Make sure that the maintainer is still fine with it and you can commit it.

I don't believe this port would even work if the user overrides WEBAPP_HOME to something other than the one from OPTIONS.

Umm, you're right! The options include is not necessary when using helpers.
Also, this works:

Note that bsd.port.options.mk is included before VARS are used.

Also, it's an ordering issue. This works:

You can triple-backquote (`) text block to paste it into Phabricator, like so:

You can still fetch it from GitHub and build it. This should be preferable to fetching precompiled jar files.

The "/var/db" should be excluded from this test. Maybe instead we should test if a port puts anything in LOCALBASE/var (as, say, prosody does) and error on that.