Remove support for FreeBSD < 10.3

[rene]

Summary

Now that FreeBSD 9.3, 10.1 and 10.2 are no longer supported by so@, remove support for these versions from Mk/* and Mk/Uses/*

Lightly tested with an INDEX-12 build.

Test Plan

'make index' seems OK

[adamw]

Mk/Uses/linux.mk

45 ↗

(On Diff #23507)

This new language became somewhat vague. The original message was an approximation, so this could be too:

Linux ${LINUX_DEFAULT} is only supported on FreeBSD/amd64 11.0-RELEASE or higher

[rene]

Mk/Uses/linux.mk

45 ↗

(On Diff #23507)

It is supported on 10.3-RELEASE, so the massage is more correct. But perhaps we can suggest users to update to 11.0-RELEASE, because they are running some unsupported 11.0-CURRENT ?

[rene]

Mk/bsd.port.mk

1637 ↗

(On Diff #23507)

I was indeed thinking about simplifying this to

.if ${OSVERSION} >= 1003503

[adamw]

Mk/Uses/linux.mk

45 ↗

(On Diff #23507)

Gotcha. I mean presumably if someone is running 11.0-CURRENT they understand what "not supported on early -CURRENT" means.

Linux ${LINUX_DEFAULT} is unsupported on pre-release versions of FreeBSD 11. Update to 11.0-RELEASE or higher

Mk/bsd.port.mk

1637 ↗

(On Diff #23507)

Yes

[rene]

• Clarify an IGNORE message after suggestion from adamw
• Simplify an if statement

[rene]

Mk/bsd.port.mk

1641 ↗

(On Diff #23512)

This else block is still valid for 10.3-RELEASE and early 10.3-STABLE, i.e. versions 1003000 to 1003502.
For 11.0-CURRENT and later this does not apply, see https://svnweb.freebsd.org/base?view=revision&revision=264661

[mat]

If you really want to remove support for FreeBSD < 10.3, it also means removing fmake support, for example, the silly .for constructs we had to use.

For example, in a loop like this, the _G_TEMP can be removed (and replaced by _group).

.       for _group in ${_S_TEMP:S/,/ /g}
_G_TEMP=    ${_group}
.           if ${_G_TEMP:C/[a-zA-Z0-9_]//g}
check-makevars::
                @${ECHO_MSG} "The ${_S} MASTER_SITES line has"
                @${ECHO_MSG} "a group with invalid characters, only use [a-zA-Z0-9_]"
                @${FALSE}
.           endif
.           if ${_G_TEMP} == all || ${_G_TEMP} == ALL || ${_G_TEMP} == default
check-makevars::
                @${ECHO_MSG} "Makefile error: the words all, ALL and default are reserved and cannot be"
                @${ECHO_MSG} "used in group definitions. Please fix your MASTER_SITES"
                @${FALSE}
.           endif
_MASTER_SITES_${_group}+=   ${_S:C@^(.*/):[^/:]+$@\1@}
.       endfor

becomes

.       for _group in ${_S_TEMP:S/,/ /g}
.           if ${_group:C/[a-zA-Z0-9_]//g}
check-makevars::
                @${ECHO_MSG} "The ${_S} MASTER_SITES line has"
                @${ECHO_MSG} "a group with invalid characters, only use [a-zA-Z0-9_]"
                @${FALSE}
.           endif
.           if ${_group} == all || ${_group} == ALL || ${_group} == default
check-makevars::
                @${ECHO_MSG} "Makefile error: the words all, ALL and default are reserved and cannot be"
                @${ECHO_MSG} "used in group definitions. Please fix your MASTER_SITES"
                @${FALSE}
.           endif
_MASTER_SITES_${_group}+=   ${_S:C@^(.*/):[^/:]+$@\1@}
.       endfor

Those constructs were necessary because fmake replaced each occurence of the for variable by the value it was currently iterating, whereas newer make created temporary variables holding the value.

[mat]

Looks good, I have not checked all the tests you removed though.