Update lang/python27 to 2.7.8, devel/py-setuptools to 5.4.1

[koobs]

Summary

lang/python27:

• Update to 2.7.8, update pkg-plist accordingly
• OPTIONS: Reorder and sort alphabetically
• OPTIONS: Make POSIX SEM(aphores) a default (Hi binary package users!)
• OPTIONS: Add better descriptions for UCS2 and UCS4
• Use OPTIONS helpers
• Remove patch-CVE-2014-1912: upstream
• Remove patch-issue20374: upstream
• Rename patch-Doclibraryfcntl.rst: underscore convention
• Rename patch-Modules__fcntlmodule.c: underscore convention
• Modify patch-setup.py: revert ncurses overrides, add partial backport for ossaudio checks
• Remove pkg-message warning for POSIX Semaphores. They are now in GENERIC for 8.x 9.x, 10.x and tested well upstream
• Tweak pkg-message
• Remove platformX bits from Makefile/pkg-plist
• Add regression-test target

devel/py-setuptools

• Update to 5.3
• Rename patch-setuptools_command_install_egg_info.py (conventions)
• Whitespace alignment

Test Plan

• https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=191405
• lang/python27: regression-test passes on 9-STABLE
• devel/py-setuptools: Ran 135 tests in 2.274s FAILED (failures=1, skipped=1) (Known Failure)
• Poudriere logs: https:// koobs dot io/poudriere/bulk/latest-per-pkg/python27/2.7.8/
• Needs exp-run

[wg]

Hi,

SEM_CONFIGURE_ON= ac_cv_posix_semaphores_enabled=yes
SEM_CONFIGURE_OFF= ac_cv_posix_semaphores_enabled=no

Shouldnt this be SEM_CONFIGURE_ENV_ON

?

[mat]

In D364#8, @wg wrote:

Hi,

SEM_CONFIGURE_ON= ac_cv_posix_semaphores_enabled=yes
SEM_CONFIGURE_OFF= ac_cv_posix_semaphores_enabled=no

Shouldnt this be SEM_CONFIGURE_ENV_ON

?

Work both in the ENV and as an argument. (Also, we don't have *_CONFIGURE_ENV_*)

[wg]

In D364#9, @mat wrote:

In D364#8, @wg wrote:

Hi,

SEM_CONFIGURE_ON= ac_cv_posix_semaphores_enabled=yes
SEM_CONFIGURE_OFF= ac_cv_posix_semaphores_enabled=no

Shouldnt this be SEM_CONFIGURE_ENV_ON

?

Work both in the ENV and as an argument. (Also, we don't have *_CONFIGURE_ENV_*)

Yes, we do, see /usr/ports/Mk/bsd.options.mk _OPTIONS_FLAGS.

There are places using CONFIGURE_ENV_ON and other places using CONFIGURE_ON, just trying to understand why the discrepancy.

[wg]

In D364#11, @wg wrote:

In D364#9, @mat wrote:

In D364#8, @wg wrote:

Hi,

SEM_CONFIGURE_ON= ac_cv_posix_semaphores_enabled=yes
SEM_CONFIGURE_OFF= ac_cv_posix_semaphores_enabled=no

Shouldnt this be SEM_CONFIGURE_ENV_ON

?

Work both in the ENV and as an argument. (Also, we don't have *_CONFIGURE_ENV_*)

Yes, we do, see /usr/ports/Mk/bsd.options.mk _OPTIONS_FLAGS.

There are places using CONFIGURE_ENV_ON and other places using CONFIGURE_ON, just trying to understand why the discrepancy.

Allow me to correct myself, there is a place using CONFIGURE_ENV_OFF not ON.

Reading the code more carefully we have OPT_CONFIGURE_ENV= and OPT_CONFIGURE_ENV_OFF=, but no OPT_CONFIGURE_ENV_ON= (which was what I proposed and seems it should actually be OPT_CONFIGURE_ENV=).

Cheers,

[bapt]

Noportdata is still in use so the hook will prevent you from committing that, btw the build procedure is still weird, can we with that change include the clean up needed, like it was done with python34

[wg]

There is still a @${CAT} ${PKGMESSAGE} as well.

[koobs]

Update: Options helpers, NOPORTDATA, PKGMESSAGE

Modify SEM_CONFIGURE_OFF -> SEM_CONFIGURE_ENV_OFF (wg@)
Remove NOPORTDATA conditional (bapt@)
Remove ${CAT} ${PKGMESSAGE} (wg@)

[koobs]

Blocked on review/accept of updated changes.

[mva]

lang/python27/Makefile

64

This should be Enable UCS4 Unicode Strings

66–67

FPECTL_CONFIGURE_WITH= fpectl

82

UCS2_CONFIGURE_ENABLE= unicode=ucs2

(not sure, if that works, but it should)

83

UCS4_CONFIGURE_ENABLE= unicode=ucs4

(not sure, if that works, but it should)

[koobs]

lang/python27/Makefile

64

Yep, good find

66–67

I tried to be super explicit, since I cant see a --without-fpectl case in configure, only a --with-fpectl=no case:

# Check for --with-fpectl
{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for --with-fpectl" >&5
$as_echo_n "checking for --with-fpectl... " >&6; }

# Check whether --with-fpectl was given.
if test "${with_fpectl+set}" = set; then :
  withval=$with_fpectl;
if test "$withval" != no
then

I know most configure's are supposed to support enable/disable/with/without/=yes/=no transparently for all if not most cases, but I've also see very fail configures that only support VERY specific syntaxen

82

As above, opt_CONFIGURE_ENABLE is nicely declarative, but has the downside of adding an implicit --disable-${content} value when the option is disabled:

# ${opt}_CONFIGURE_ENABLE       Will add to CONFIGURE_ARGS:
#                               Option enabled  --enable-${content}
#                               Option disabled --disable-${content}

83

As above, opt_CONFIGURE_ENABLE is nicely declarative, but has the downside of adding an implicit --disable-${content} value when the option is disabled:

# ${opt}_CONFIGURE_ENABLE       Will add to CONFIGURE_ARGS:
#                               Option enabled  --enable-${content}
#                               Option disabled --disable-${content}

[mat]

lang/python27/Makefile

82

nope, it only works with CONFIGURE_WITH, because you can have say --enable-codec=mp3 --disable-codec=x264 or something (at least, that's what I was told when I asked why not add it to _CONFIGURE_ENABLE too.)

[koobs]

Fix typo in UCS4 option description (mva@)

[mat]

lang/python27/Makefile

82

see http://www.freebsd.org/doc/en/books/porters-handbook/makefile-options.html#options-configure_enable

[mat]

lang/python27/Makefile

82

Also, in that case, it should *NOT* be _CONFIGURE_ENABLE, but only _CONFIGURE_ON.

[mva]

lang/python27/Makefile

66–67

Okay, agreed - we should add a comment then, so that others do not "clean it up" and possibly break behaviour.

#Using _ON instead of _WITH here, since --without-fpectl might not be explicitly handled
FPECTL_CONFIGURE_ON=  --with-fpectl

[koobs]

In D364#34, @mva wrote:

Okay, agreed - we should add a comment then, so that others do not "clean it up" and possibly break behaviour.

#Using _ON instead of _WITH here, since --without-fpectl might not be explicitly handled
FPECTL_CONFIGURE_ON=  --with-fpectl

Perhaps those notes are more useful in port.options.mk

[mva]

In D364#35, @koobs wrote:

In D364#34, @mva wrote:

Okay, agreed - we should add a comment then, so that others do not "clean it up" and possibly break behaviour.

#Using _ON instead of _WITH here, since --without-fpectl might not be explicitly handled
FPECTL_CONFIGURE_ON=  --with-fpectl

Perhaps those notes are more useful in port.options.mk

No. Usually the autotools magic for with/without works just fine. If you noticed a different behaviour, that would be most likely specific for the script in question (until proven otherwise ;-).

[bapt]

I carefully closed my eyes on the static/share thing (which I still think should be cleaned out in the same time because it is blocking my work on cross building :))

I still have a couple of others comments
Given pkg-message does not have anymore any pattern to be changes by SUB_LIST it is useless to still name it .in leave it in files, that makes run a complicated sed for nothing.

Others comments are inlined

lang/python27/Makefile

85

I see nothing that requires .bsd.port.pre.mk why not just bsd.port.options.mk?

179

MSEM_EXTRA_PATCHES= ${FILESDIR}/extra-patch-setup.py
Should do the above in a nicer way and with better control

207

Useless use of cat here :) ${AWK} ... ${PLIST_TEMPLATE} would be simpler :)

250

let's in case earlier pre.mk has been replaced by options.mk replace that by bsd.port.mk

[mva]

And now it messes up some builds for people using the PTH option. It seems to be misleading for people and should be removed. See

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=189844
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=191744

for more details.

[koobs]

I'd like to decouple any PTH changes from this commit, as it has already been delayed and it clearly needs further investigation independent from the 2.7.8 update.

Also, the NOPORTDATA change has now been committed by @sbz in https://svnweb.freebsd.org/ports?view=revision&revision=361574

lang/python27/Makefile

85

Ack'd

179

The SEM option will be removed along with PTH (and potentially others) in subsequent changes, with associated cleanups/upstreaming that will be possible. Will not include any SEM options/mechanism modifications in this 2.7.8 update

207

Not confident enough to change this given I don't know what it does, or achieves. Can you provide the alternative code?

250

Ack'd

[koobs]

pkg-message, SUB_FILES and bsd.port.options.mk

• files/pkg-message no longer needs substitutions, remove .in suffix and SUB_FILES (bapt@)

• Replace bsd.port.{pre,post}.mk with port.options.mk (unecessary) (bapt@)