PHP Flavors.

[mat]

rebase.

[mat]

• Uniquify pecl package names.
• And pear.
• Don't do pearXY- and peclXY-, do phpXY-pear/pecl-
• force php flavors for pear/horde, and provide multiple pkgnameprefixes
• Add pear:env.
• Fix pear ports.
• Flavorize the pear-channel ports.
• Handle pear channels dependency automatically.
• poudriere bulk */pecl-* */pear-* */horde-* starts.

[tz]

It looks fine too me. I'm glad that i wasn't too far away from the Flavor code. :)

After some tests i wouldn't extend this patch further - its great enough. I would flavorize various PHP ports afterwards. :)

[joneum]

why only www/wordpress?
There are more wordpress ports:
russian/wordpress
and more here: https://portscout.freebsd.org/[e-mail]

Please node: wordpress 4.9.3 is out and i will update it (bugfix release)

[mat]

• Add flavors.
• more ports

[mat]

• Allow using pecl with github.
• Always define PECL_PKGNAMEPREFIX.
• Pass the flavor to external pecl extensions.
• More flavors.

[mat]

In D14208#298316, @tz wrote:

It looks fine too me. I'm glad that i wasn't too far away from the Flavor code. :)

After some tests i wouldn't extend this patch further - its great enough. I would flavorize various PHP ports afterwards. :)

Would be nice if it was a good idea, but for many it is not.

In D14208#298437, @joneum wrote:

why only www/wordpress?
There are more wordpress ports:
russian/wordpress
and more here: https://portscout.freebsd.org/[e-mail]

Please node: wordpress 4.9.3 is out and i will update it (bugfix release)

Why only www/wordpress? Because it's the first PHP app that came to mind, and I wanted to toy with it.

[mat]

• more ports.
• bulk -n -a passes

[tz]

In D14208#298635, @mat wrote:

In D14208#298316, @tz wrote:

It looks fine too me. I'm glad that i wasn't too far away from the Flavor code. :)

After some tests i wouldn't extend this patch further - its great enough. I would flavorize various PHP ports afterwards. :)

Would be nice if it was a good idea, but for many it is not.

I don't know who you mean? Currently there are only 3 persons (+ portmgr) in this review.

I won't stop you. But the patch exceeds my review-time and touches many different issues. If you don't want to break it down, please go ahead, but in this case i'm out. Maybe joneum can help with more time. And it would be great if this feature enters the portstree.

But one note: please add more comments to the framework-changes. It is already hard to understand what the framework does. And with the current patch it becomes harder.

[mat]

In D14208#298688, @tz wrote:

In D14208#298635, @mat wrote:

In D14208#298316, @tz wrote:

It looks fine too me. I'm glad that i wasn't too far away from the Flavor code. :)

After some tests i wouldn't extend this patch further - its great enough. I would flavorize various PHP ports afterwards. :)

Would be nice if it was a good idea, but for many it is not.

I don't know who you mean? Currently there are only 3 persons (+ portmgr) in this review.

Sorry, my sentence was less than clear. The last "it" refered to "flavorizing php ports". There are many that cannot be done afterwards.

I won't stop you. But the patch exceeds my review-time and touches many different issues. If you don't want to break it down, please go ahead, but in this case i'm out. Maybe joneum can help with more time. And it would be great if this feature enters the portstree.

The only interesting bits are the Mk changes, the rest are just making things build.

But one note: please add more comments to the framework-changes. It is already hard to understand what the framework does. And with the current patch it becomes harder.

I usually add comments later, when the dust settles a bit, because there's no point in spending time explaining stuff that will be rewritten five times during the process of the review.

[mat]

rebase.

[mat]

Rebase +

• Allow the installed php version to be the default flavor.
• Remove flavors from PHP apps.

[mat]

rebase after fixing devel/pear-channel-htmlpurifier.org

[mat]

• Fail early if the installed PHP version is not supported.

[mat]

rebase and

• Add a few comments.

[tz]

So whats the current status here? When will it land?

[mat]

In D14208#302810, @tz wrote:

So whats the current status here?

It is ready, been ready for a while, waiting for review.

When will it land?

When I decide I am tired of waiting for a review, and after an exp-run dance :-)

[tz]

Great - i will try to find some time within the next days for review :)

[mat]

rebase, and:

• Merge databases/php7*-memcache
• Merge ftp/php*-fastdfs
• Remove PKGNAMESUFFIX for flavorized PHP extensions when they different
• Flavorize www/unit-php*

[tz]

Fine - this is only a rough review. I found some issues, have some questions and many entries reflecting that i'm unsure what you do. :D But hopefully this helps.

MOVED

9968

Merge conflict?

9986

The 8 lines above: typo. Must be "Merge" instead of "Metge"

Mk/Uses/horde.mk

18

I don't understand why this can be removed without any change. Is this channel no longer needed?

Mk/Uses/php.mk

113

I don't understand the new conditional. Why is it used here?

203

What is WRT? Also "dark magic" is not a helpful documentation

databases/Makefile

634

php70-memcached needs to be flavorized too.

databases/pear-DoctrineCommon/Makefile

15

Why could this be removed without any replacement?

databases/pear-Horde_Mongo/Makefile

10–11

No need to update a comment. Better remove it.

databases/pecl-memcached2/Makefile

8

I'm not sure- why can this be thrown away?

databases/php-memcache/pkg-descr

8

This is not correct.

databases/php-memcached/Makefile

1

This port should be renamed to php-memcached after flavorizing.

devel/pecl-jsmin2/Makefile

7

Won't this conflict with devel/pecl-jsmin?

devel/pecl-msgpack0/Makefile

7

Conflict with devel/pecl-msgpack?

devel/pecl-propro2/Makefile

8

Conflict with devel/pecl-propro?

devel/pecl-weakref2/Makefile

7

Conflict with devel/pecl-weakref?

mail/pecl-mailparse2/Makefile

8

Conflict with mail/pecl-mailparse?

math/pecl-bitset2/Makefile

7

Conflict with math/pecl-bitset?

math/pecl-stats2/Makefile

6

Conflict with math/pecl-stats?
Okay... i'm skipping this types of questions from here on. Its a little redundant (especially if i'm wrong ;D)

[mat]

Note that the only interesting bits to review are the Mk/ changes, all the rest is mechanical changes to make sure everything still work.

The PKGNAMESUFFIX are removed in a few ports because they do not conflict any more.

Mk/Uses/horde.mk

18

It is done automatically by pear.mk

Mk/Uses/php.mk

113

Because if a pecl port USE_GITHUB to fetch their distfile, they can still USES=php:pecl to get the right suffix and DIST_SUBDIR.

databases/Makefile

634

it is.

databases/pear-DoctrineCommon/Makefile

15

pear.mk

databases/php-memcached/Makefile

1

it could, yes.

[mat]

• Fixes after tz's review.

[mat]

About the PKGNAMESUFFIX removals, right now, we have:

$ for i in devel/pecl-weakref{,2}; do make -C $i pretty-flavors-package-names; done
no flavor: pecl-weakref-0.2.6_1
no flavor: pecl-weakref2-0.3.3

So the 2 is needed.

With flavors, we have:

$ for i in devel/pecl-weakref{,2}; do make -C $i pretty-flavors-package-names; done
php56: php56-pecl-weakref-0.2.6_1
php70: php70-pecl-weakref-0.3.3
php71: php71-pecl-weakref-0.3.3
php72: php72-pecl-weakref-0.3.3

So the 2 is no longer needed, package names are already different.

[tz]

Besides the 2 cosmetic issues i'm fine with it (as far as i understand it).

I'm looking forward to see it getting committed! :)

Mk/Uses/php.mk

203

For the log: i'm insisting that "dark magic" is not a helpful documentation at all.

databases/pear-Horde_Mongo/Makefile

10–11

The issue isn't done? The line is still a comment without any functionality?

[mat]

rebase.

[tz]

Issues were addressed, looks fine to me! :)
Thanks!

[mat]

• Maj.

[mat]

rebase.

[mat]

• Add CHANGES and UPDATING entries.

[mat]

• reword a bit.
• Add a bit about package name changes.

[tz]

This was already committed and could now be abandoned?