Spelling: 'component'

further more this comment contradicts the statement later on the USES=gl takes no arguments.

Could you maybe expand component and argument a bit, as just afterwards yo say the script does not take any.

Remove quotes around yes.

USE_GL=yes should generate a DEV_WARNING too, as this is now only an optional argument to the USES.

This whole bit should be moved to Mk/Uses/gl.mk.

You should add a check testing that USES does not already contains gl.

Also, add a DEV_WARNING saying that using USE_GL without USES=gl is wrong.

This should only be a compat shim, so it should only add gl if it is not already in USES. It should also add a DEV_WARNING saying that it is invalid to use USE_GL without USES=gl.

Ok, however, USE_GL= is not a compat shim, it's still needed to specify which GL components to bring in. But perhaps I'm misunderstanding what you mean?

Thank you for your input, I worded it wrong.

It should only add if USES does not include gl. Then it needs to have a warning like this:

DEV_WARNINGS+=   "Using USE_GL alone is deprecated, please add USES=gl."

Thank you very much for the clarification.

s/gl/glu/

This should be glu too.

Doh. Will fix.

No, this is the case when USES=gl but USE_GL is not defined at all, then we should use gl. The case when USE_GL is yes is handled below, and defaults to glu, just like before.

Having

1. USE_GL=yes -> USE_GL=glu
2. USE_GL is not defined -> USE_GL=gl

is really really a bad idea. yes and not defined should mean the same thing.

I do not agree.
No port today has USES=gl and USE_GL not defined, so there is no backwards compatibility to take into account. I understand USE_GL=yes means glu, that's because it's been that way for a very long time. USES=gl is new, and can imply new things. It would surprise me as a developer that USES=gl actually means i get glu, not gl unless I specifically asked for glu with USE_GL.

Should say something about having to set USE_GL.

Why move this?

Noted, will fix.

To fix the build. USES need to be defined before bsd.port.pre.mk is included, and the port has USE_GL at line 42, which implies USES. See build failure log here:
http://package23.nyi.freebsd.org/data/111i386-default-PR230692PR230909/2018-08-29_18h58m56s/logs/errors/q3cellshading-1.0_3.log and comment from Antoine here: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=230692#c6

Yeah, so, you changed the include to be bsd.port.options.mk, I get that, but why move this block, it was fine up there.

Because it had to be after the bsd.port.pre.mk include, otherwise it wouldn't build..

Just changing the include from bsd.port.pre.mk gives me this:

make: "/usr/ports/games/q3cellshading/Makefile" line 91: Malformed conditional (${COMPILER_TYPE} == clang)
make: "/usr/ports/Mk/bsd.port.mk" line 1834: Malformed conditional (${WITH_PKG} == devel)
make: "/usr/ports/Mk/bsd.port.mk" line 2140: Malformed conditional (${PREFIX} == /usr)
make: "/usr/ports/Mk/bsd.port.mk" line 2658: Malformed conditional (${PREFIX} == /usr)
make: "/usr/ports/Mk/bsd.port.mk" line 2671: Malformed conditional (${PREFIX} == /usr)
make: "/usr/ports/Mk/bsd.port.mk" line 4553: Malformed conditional ((${PREFIX} != ${LOCALBASE} && ${PREFIX} != ${LINUXBASE} && ${PREFIX} != "/usr" && ${PREFIX} != "/" && !defined(NO_PREFIX_RMDIR)))
make: Fatal errors encountered -- cannot continue
make: stopped in /usr/ports/games/q3cellshading

So bsd.port.pre.mk needs to be included before the block, but after the uses stuff, which is in the options block.

from bsd.port.pre.mk to bsd.port.options.mk

This looks more like a symtom of not changing bsd.port.post.mk to bsd.port.mk at the end for the Makefile. So my point stands.

Even changing the bsd.port.pre.mk to bsd.port.options.mk and bsd.port.post.mk to bsd.port.mk I still get the following error:

make: "/usr/ports/games/q3cellshading/Makefile" line 90: Malformed conditional (${COMPILER_TYPE} == clang)
make: Fatal errors encountered -- cannot continue

Adding bsd.port.pre.mk here is fine, but you do not need to move the rest.