Re: PHP 5.6 life cycle

On 12/06/2015 11:36 AM, Zeev Suraski wrote:

-----Original Message----- From: Nikita Popov [mailto:nikita.ppv@gmail.com] Sent: Sunday, December 06, 2015 7:03 PM To: Ferenc Kovacs Cc: Jan Ehrhardt; PHP Internals Subject: Re: [PHP-DEV] PHP 5.6 life cycle On Sun, Dec 6, 2015 at 5:32 PM, Ferenc Kovacs <tyra3l@gmail.com> wrote:

2015. dec. 6. 13:15 ezt írta ("Jan Ehrhardt" <phpdev@ehrhardt.nl>):

See http://php.net/supported-versions.php Will PHP 5.6 go into 'security fixes only' on 28 Aug 2015 with a end of life on 28 Aug 2016? Or will we be postponing this a couple of months? BTW: An end-of-life in Dec 2016 will be in line wih the EOL of OpenSSL 1.0.1: "Version 1.0.1 will be supported until 2016-12-31." http://openssl.org/policies/releasestrat.html -- Jan -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php

Since the rfc for 5.7 failed the voting I've personally assumed that we don't want to support the 5.x series after the normal lifecycle for 5.6: https://wiki.php.net/rfc/php57 Most of my arguments for 5.7 was the same as Zeev and orhers listed here in this thread but the majority shared the opinion that the support left for 5.6 is sufficient and we shouldn't prolong the support for 5.x as it will just delay the adoption for 7.0

I can't say anything as to what majorities think, but while I did not want a PHP 5.7 release, with the large amount of additional work and fragmentation of focus it would have implied, this does not make me adverse to extending the PHP 5.6 support cycle. I would go as far as saying that us not having done a PHP 5.7 release is an argument in favor of prolonging support for PHP 5.6, not the reverse.

I agree completely. Ferenc - the way I see it, 5.7 actually had little to do with the arguments I brought up. I believe that the main reason 5.7 was opposed (at least I can at least speak for myself) is that people felt it wasn't a good idea to divide our attention from delivering 7.0, something that 5.7 (even if the only new features were forward compatibility, and more realistically - packing extra features) would have done. Sebastian - while it's obvious that us supporting PHP 5.6 for a while longer does have some effect on migration to 7.0, realistically, we can't force millions of people to upgrade on our own timeline if it's too short. On such a short timeline, what it practically means is that there are going to be a lot more websites that won't migrate on time and will become insecure on September 2017. Also, discontinuing support for PHP 5.6 in August means you'd have less time to migrate from 5.6 to 7.0 than you did to upgrade from 5.5 to 5.6 - and that was a painless upgrade. What if 7.0 was delayed by a few more months? Or a year? Both seemed like likely scenarios back when we called the 7.0 timeline aggressive. The 'sin' of the PHP 4 EOL was - well - that we didn't have one for a very long time. We should definitely have a clear one for 5.6, but it should be more realistic than 1.5yrs away. In general, I don't think timelines make sense to commit to before a version is released. If for whatever reason a release gets delayed it makes no sense that you'd be forced, as a user, for a shorter upgrade cycle. Something along the lines of Francois' suggestion - where the lifetime of version N-1 relates to the release date of version N is definitely needed, and that was the thinking behind the release process timeline to begin with. Zeev

Coming from user-space, +1 to the "an extra year over whatever it would have had" for the last release within a major. While I'm champing at the bit to use PHP 7 and will be trying to push others to do the same, realistically it is a bumpier upgrade than 5.5->5.6 (duh) so giving people extra breathing room to plan an upgrade is a good thing. Whether that's an extra year of active support or security support, I don't feel strongly. (Maybe making it security support will be a better upgrade incentive?) That said, I also agree that a fixed date is mandatory, or people won't feel any pressure. The drop-dead date is important for planning and messaging, just as we saw with GoPHP5. That extra year is also helpful to all the big projects that just increased their minimum version to 5.5. (Drupal, Symfony, Zend Framework, Guzzle, etc.) I expect most projects are going to skip 5.6 as a minimum required version and jump straight to 7, just like many skipped 5.4, but they'll need to see more server offerings. It will be an easier case to make in 2 years for the next version of all of those projects to go to 7.x if 5.6 is sec-only with a known drop-dead date, but not entirely abandoned so we have some breathing room until then. --Larry Garfield