Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:105408
MIME-Version: 1.0
References: <CAFPFaMJEn=gjk5jAAxR2bF=4Mt3r0FJCm6mbOCReOmyybtAkkg@mail.gmail.com>
 <CAOv67gsiva80KEhw7D2N2qKCz0gZvza30EDL7SZoJRPFsGoROA@mail.gmail.com>
 <0ec42fa9-77d1-a203-8425-e72fdd5071f3@korulczyk.pl> <06473788-a34b-f041-36e6-31d19d8dda4c@cubiclesoft.com>
 <59cafbfb-2bb0-468c-458f-74bcac780e0f@korulczyk.pl> <004c01d4f09f$880ac320$98204960$@roze.lv>
 <CAFPFaM+QwXpm-mfwRtbT2sPB_C-9wvPELVTV5dHvizgmfUjMnw@mail.gmail.com>
 <004401d4faa3$60f83700$22e8a500$@gmail.com> <2f922f17-bc7c-313a-8f77-122e861995be@lsces.co.uk>
 <5741936F-B1F4-43C7-B815-F9D8030AC7BB@koalephant.com> <e5c10f59-bde9-dadc-e837-266f870b6358@lsces.co.uk>
 <B392E7C8-BBAC-48D2-BBBA-3D7C7EFD9595@koalephant.com> <e702a5cc-9567-0591-8f5e-b14aef0766b7@lsces.co.uk>
 <CAAfnsFU1B61qaPRAh5uNTWMALcV5xmae0Yb1yY9+QaAGLVGx7A@mail.gmail.com>
 <49A4B76C-4C62-4CBE-BA20-FBE56CA29AB0@cschneid.com> <CADyq6sL7m=A=pA1rNBg_CFOoMtfCZn50AfALh9jtLHvn9NRzmw@mail.gmail.com>
 <609E93CF-099B-446C-AD28-04F1D802C9F0@cschneid.com> <CADyq6s+6Gqaa6Werc0Vn=gKWrrqC71+ry1rpsPEpq4BSE4wq_Q@mail.gmail.com>
 <CAPKYkKz4u-D3ebVJCU8TgCuEGVVC28-YSj76sCU1TS4BaDjeBA@mail.gmail.com>
In-Reply-To: <CAPKYkKz4u-D3ebVJCU8TgCuEGVVC28-YSj76sCU1TS4BaDjeBA@mail.gmail.com>
Date: Wed, 24 Apr 2019 21:07:19 +0200
Message-ID: <CAAfnsFUAY_O9JbDWMsntFKz1otTjE0BKYrsMif3DzG+zQpahxw@mail.gmail.com>
To: Chase Peeler <chasepeeler@gmail.com>
Cc: Marco Pivetta <ocramius@gmail.com>, Christian Schneider <cschneid@cschneid.com>, 
	PHP Internals List <internals@lists.php.net>
Content-Type: text/plain; charset="UTF-8"
Subject: Re: [PHP-DEV] [RFC] [VOTE] Deprecate PHP's short open tags
From: peterkokot@gmail.com (Peter Kokot)

Hello,

On Wed, 24 Apr 2019 at 19:44, Chase Peeler <chasepeeler@gmail.com> wrote:
>
> On Wed, Apr 24, 2019 at 1:27 PM Marco Pivetta <ocramius@gmail.com> wrote:
>
> > On Wed, 24 Apr 2019, 19:25 Christian Schneider, <cschneid@cschneid.com>
> > wrote:
> >
> > > Am 24.04.2019 um 19:13 schrieb Marco Pivetta <ocramius@gmail.com>:
> > > > On Wed, 24 Apr 2019, 19:10 Christian Schneider, <cschneid@cschneid.com
> > >
> > > wrote:
> > > > Am 24.04.2019 um 19:01 schrieb Peter Kokot <peterkokot@gmail.com>:
> > > > > just a friendly reminder that by the time one writes an email here
> > > > > these tags can be already replaced with the usual ones.
> > > >
> > > > A friendly reminder that some people are hosting customer code which
> > > they do not want to touch but will get support requests once the code
> > > breaks.
> > > >
> > > > - Chris
> > > >
> > > > That's normal? Everyone has projects to maintain, and breaking changes
> > > are common: they're gonna call you for one anyway: if you don't like
> > that,
> > > then you are in the wrong line of business.
> > >
> > > See Chase Peeler's point: A breaking change should have a reward big
> > > enough to justify it.
> > > And that's what where we (including Zeev Suraski and other core
> > > developers) disagree.
> > >
> > > - Chris
> > >
> >
> > Run a fixer: they are out there, and they are extremely stable too.
> >
> > Also a good chance to finally take a look at code that has been rotting in
> > a hard drive for too much time.
> >
> All of that takes time though. I have 6,787 short opening tags found. Even
> if I use a fixer to generate a diff, or, to fix them and then examine the
> diff in a pull request... that's going to take a LOT of time. It's going to
> start getting messy if I find false positives and need to exclude changes.
> It still doesn't address the impact of changes that aren't found. Are you
> 100% positive that the fixers out there will catch EVERY single instance?
> php-cs-fixer doesn't update <? inside of strings. If that's generating
> dynamic code, then that's not getting caught.
>
> The output from php-cs-fixer just generated a diff file that was 161,000
> lines long. But yea, that's something I can scan through real quick and
> make sure nothing is going to get broken. No chance I'll miss something
> either.
>
> I would LOVE to have the time to go through our legacy code and clean out
> old stuff. We have a lot of technical debt that, while we aren't paying it
> off at the moment, it's gaining any interest either. We also have plenty
> that is. It's tough to justify putting off the stuff that is gaining
> interest to focus on stuff that isn't so we can fix short tags.
>
> I don't work for a software company. I develop an internal application for
> a non-software company. There are 2 other developers and enough work for 10
> developers. It's going to be VERY hard for me to get management to approve
> putting off projects that have a direct impact on our business in order to
> upgrade to PHP 8 when that time comes. I think you are going to find that's
> a pretty common occurrence, and the adoption rate of PHP 8 is going to be
> VERY low. Especially when more and more user-friendly alternatives like
> python and node are coming along. It was one thing when the options were
> RoR, ASP.NET, and JSP. That's not the ecosystem anymore, and it's only
> going to provide additional user friendly opportunities in the next couple
> of years leading up to PHP8.
>
> Can someone PLEASE tell me the positive gains this RFC will accomplish that
> justifies risking:
> 1.) Losing market share
> 2.) losing credibility
> 3.) removing a large number of libraries that are fine from a technical
> aspect, but will stop working due to the existence of <?
> 4.) new security risks (code leaks that are more likely than the code leaks
> the RFC seeks to prevent)
>
> And, please don't use the existence of code fixers to justify this unless
> you are willing to demonstrate how it's quick and easy to go through a
> 161,000 line diff file or are willing to 100% guarantee that the fixer will
> not break anything, will not fix anything it shouldn't, and will not miss
> anything it should fix.
>
>
>
>
>
> --
> Chase Peeler
> chasepeeler@gmail.com

I've done few commits with about 8k files changed with very repeating
changes like this and without breaking anything. It will take you
about 30minutes... Let's say, one hour for also taking a break from
all the scrolling the git diff and to return the merge the next day
after another check or something like this.

The change (speaking for 8k files using short opening tags converted
to <?php) itself takes about 80 seconds with using PHP CS Fixer but
the memory_limit needs to be increased a bit and only relevant change
is done in all files. 8k (or 6k) opening tags is nothing very big
actually.

Then also search for similar occurrences by using a regex for those
edge cases you're mentioning in strings and such (I doubt there are
but ok) and you have this fixed. Then run this app on PHP 8 and you'll
see the other BC breaks that happen due to the major release. Major
releases are meant to break things because otherwise we can only load
the language with new things and maintain old stuff in there for
eternity.

That's why serious apps have support cycles and follow their
frameworks release and life cycle, and in the end also PHP cycle.
Simple as that.

--
Peter Kokot