# PHP 5.3.0RC3

- Thread: https://externals.io/message/43869
- Original archive: https://news-web.php.net/php.internals/43869
- Started by Lukas Kahwe Smith on 2009-05-08T08:31:21+00:00
- Messages: 12
- Reply by email to internals@lists.php.net
- More for agents: search with https://externals.io/search.md?q={query} · JSON API at https://externals.io/api · documentation at https://externals.io/llms.txt

---

## #43869 — Lukas Kahwe Smith — 2009-05-08T08:31:21+00:00

URL: https://externals.io/message/43869#43869

Aloha,

We have finally released RC2. Its now time to really finish up and  
bring 5.3.0 to a stable release. I have updated the wiki page to the  
best of my knowledge [1]. If anything is missing, already done etc.  
please let me know or fix it yourself. Again the goal is to have the  
critical issues fixed ASAP, so that we can release RC3 in 2-3 weeks  
time.

Here is the list, again check the wiki page if you are interested as  
there are links pointing to addition infos for some of the issues:
1. fix open PHP 5.3 bugs
2. finish UPGRADING README file (Steph)
3. write migration guide for the manual (Steph)
4. Fix static build of extension on windows when static is the default  
and –enable-snapshot-build is used
5. Fix the NEWS file, as it skips from 5.2.4 to 5.3.0
6. `opendir()` fails on Windows directories with parent directory  
unaccessible
7. very important for APC, about caching (stat cache or other) on win/ 
fcgi (Pierre)
8. add missing libs (gmp, firebird ..) (Pierre)
9. Add `E_DEPRECATED` if session.bug_compat_42/warn is On (already  
removed in HEAD)
10. fix re2c issues: bus error, 8k script parse error
11. update short_open_tag descriptions in php.ini-* as per this (philip)

Critical issues:
1) I assume the issues with rounding are resolved. If any issues pop  
up again, please let the list know.
2) The re2c EOF bug needs a clean fix

All the while we do not need a final upgrading and migration guide for  
RC3, but the sooner the better of course. We do need them for the  
final release, which hopefully come shortly after RC3. In terms of the  
windows issues, we RMs feel that these issues existed in previous  
versions of PHP and given the amount of improvements we have for  
Windows users, we do not feel that its critical to wait for their  
resolution. That is not to say of course that getting them resolved is  
not important. However only if they are low risk changes. Most of the  
other issues are minor and should get fixed for RC3.

regards,
Lukas Kahwe Smith
mls@pooteeweet.org

[1] http://wiki.php.net/todo/php53#rc3

---

## #43882 — Lukas Kahwe Smith — 2009-05-11T12:05:57+00:00

URL: https://externals.io/message/43869#43882 · In reply to: #43869

> 1. fix open PHP 5.3 bugs


To spare us surprises so late in the game, please only commit bug  
fixes with accompanying tests. Again as Johannes mentioned, tests in  
the testfest repo do not count. They should go into php-src along with  
the actual fix.

> 2. finish UPGRADING README file (Steph)
> 3. write migration guide for the manual (Steph)

@Steph: Where do we stand here? This item needs to be in full swing to  
be finished right now. Preferably before RC3, so that this release can  
also put the documentation to the test. If you need additional hands  
or cannot complete this by this weekend alone, we need others to step  
up and help out.

> Critical issues:
> 1) I assume the issues with rounding are resolved. If any issues pop  
> up again, please let the list know.

@Matt/Dmitry: Can you just give us the quick nod that all is well here?

> 2) The re2c EOF bug needs a clean fix


@Nuno/Macus/The rest of the world: This issue is still open. If we can  
find a fix for this we are all comfortable with by Thursday evening, I  
would like to go ahead with RC3 next week. The idea being that this  
change needs to be validated a bit before its worth going into release  
mode and also re2c needs to put this change into a release as well. If  
not, then it just means we need to wait an extra week before we can  
look at releasing RC3 again.

We have identified another critical issue on windows. Pierre is in the  
process of validating the fix:
http://bugs.php.net/bug.php?id=44859

regards,
Lukas Kahwe Smith
mls@pooteeweet.org

---

## #43888 — Matt Wilmas — 2009-05-12T02:20:20+00:00

URL: https://externals.io/message/43869#43888 · In reply to: #43882

Hi Lukas,

----- Original Message -----
From: "Lukas Kahwe Smith"
Sent: Monday, May 11, 2009

>
> [...]
> > […]
>
> @Matt/Dmitry: Can you just give us the quick nod that all is well here?

No, can't say that all is well. :-/  Nothing was changed yet, sorry if you 
misunderstood.  (BTW, it's not rounding/parsing, but conversion/casting of 
floats->integers... :-))

I sent the updated patch a month ago, and then the next week Stas asked some 
questions off-list, for clarification, etc. and then said that the patch 
"looks good" and "since it appears to fix things I think it can be applied." 
That's the only feedback I had really, and Dmitry mentioned that it "breaks" 
about 30 tests (I'd consider them broken now, to match the code, however 
;-)), which I knew would have to be updated.  I didn't try to fix them yet, 
since I didn't know if the changes would finally be applied or not.  I was 
going to bring it up again but then it was too close to RC2.

There were some e-mails on the subject that I didn't follow up on (nothing 
major, just comments), including one of yours I think.  Anyway, I guess I/we 
can wonder about RC3 now?  Again, the *very minor* modifications only help 
to ensure the [usual] long-standing behavior on all platforms -- e.g. most 
users would see no change from 5.2 or prior.

I'll try to be sure to do what I can to take care of anything now, since I 
shouldn't be distracted with other stuff like leading up to RC2...

> [...]
> regards,
> Lukas Kahwe Smith
> mls@pooteeweet.org

- Matt

---

## #43889 — Jani Taskinen — 2009-05-12T02:37:09+00:00

URL: https://externals.io/message/43869#43889 · In reply to: #43888

Before committing anything else into PHP_5_3, can you first make sure everything 
  you provided is ALSO IN HEAD?!

Hint: zend_operators.*

--Jani

Matt Wilmas kirjoitti:
> Hi Lukas,
> 
> ----- Original Message -----
> From: "Lukas Kahwe Smith"
> Sent: Monday, May 11, 2009
> 
> > […]
> 
> No, can't say that all is well. :-/  Nothing was changed yet, sorry if 
> you misunderstood.  (BTW, it's not rounding/parsing, but 
> conversion/casting of floats->integers... :-))
> 
> I sent the updated patch a month ago, and then the next week Stas asked 
> some questions off-list, for clarification, etc. and then said that the 
> patch "looks good" and "since it appears to fix things I think it can be 
> applied." That's the only feedback I had really, and Dmitry mentioned 
> that it "breaks" about 30 tests (I'd consider them broken now, to match 
> the code, however ;-)), which I knew would have to be updated.  I didn't 
> try to fix them yet, since I didn't know if the changes would finally be 
> applied or not.  I was going to bring it up again but then it was too 
> close to RC2.
> 
> There were some e-mails on the subject that I didn't follow up on 
> (nothing major, just comments), including one of yours I think.  Anyway, 
> I guess I/we can wonder about RC3 now?  Again, the *very minor* 
> modifications only help to ensure the [usual] long-standing behavior on 
> all platforms -- e.g. most users would see no change from 5.2 or prior.
> 
> I'll try to be sure to do what I can to take care of anything now, since 
> I shouldn't be distracted with other stuff like leading up to RC2...
> 
> > […]
> 
> - Matt

---

## #43890 — Matt Wilmas — 2009-05-12T03:18:40+00:00

URL: https://externals.io/message/43869#43890 · In reply to: #43889

Hi Jani,

----- Original Message -----
From: "Jani Taskinen"
Sent: Monday, May 11, 2009

> Before committing anything else into PHP_5_3, can you first make sure 
> everything you provided is ALSO IN HEAD?!

To answer that directly, no. :-)  Everything I do is in HEAD first, of 
course, to keep things correct and in sync.  But I'm not going to go and 
update HEAD on behalf of someone else right at this time, which has been 
missing forever, when things are trying to be finished up for 5.3.

> Hint: zend_operators.*

(I noticed your recent cleanup stuff, and thought "He has to notice.")

I'm very well aware of anything I was involved with that has or hasn't been 
done (no hints needed). ;-)  Like many, many, many things that are (I don't 
think it's "were?") out-of-sync, Ilia can take credit for that!  Off-list 
message from him Jan 18, 2007, after asking about when HEAD would be 
updated: "The PHP 6 patches are up to Andrei to approve and commit, I am 
only looking over the 5.2 tree and occasionally the 4.4 tree as well."  That 
was 16 months before I had a CVS account...  The patch is still there [1], 
unchanged from that day, if someone wants to take it, update and apply or 
whatever before I get to it sometime after 5.3-critical stuff, I guess.

I *despise* these screwed up commits to wrong branches, so you'll never 
catch me doing one.  And I like bunnies too much (actually, don't care much 
either way, but that sounds nice. :O)).  I'd vote for karma removal until 
people can learn to do stuff correctly; you'd probably agree.

[1] http://realplain.com/php/is_numeric.diff

> --Jani

- Matt


> Matt Wilmas kirjoitti:
> > […]

---

## #43891 — Jani Taskinen — 2009-05-12T06:02:34+00:00

URL: https://externals.io/message/43869#43891 · In reply to: #43890

Matt Wilmas wrote:
> > […]
> (I noticed your recent cleanup stuff, and thought "He has to notice.")

I forgot smiley from that. This was more tongue-in-cheek than a real rant. :D

> much either way, but that sounds nice. :O)).  I'd vote for karma removal 
> until people can learn to do stuff correctly; you'd probably agree.

Yes please, but I do want to give chance for people to correct their mistakes 
before that happens. ;)

> [1] http://realplain.com/php/is_numeric.diff

Yup, that's still missing from HEAD. I'll merge it myself if nobody beats me to 
it, but I hope you could review the code afterwards so we can be sure it got in 
clean. One can't really rely on the tests on this since there are so many tests 
failing in HEAD at the moment..

Anyway, everyone: PLEASE commit to HEAD first. I know it's painful and not 
always that easy, but in the long run it makes all our lives much easier by 
making applying patches between branches actually possible. :)

--Jani

---

## #43898 — Lukas Kahwe Smith — 2009-05-12T10:58:48+00:00

URL: https://externals.io/message/43869#43898 · In reply to: #43888

> Hi Lukas,
>
> ----- Original Message -----
> From: "Lukas Kahwe Smith"
> Sent: Monday, May 11, 2009
>
> > […]
>
> No, can't say that all is well. :-/  Nothing was changed yet, sorry  
> if you misunderstood.  (BTW, it's not rounding/parsing, but  
> conversion/casting of floats->integers... :-))

Doh. Would have been nice if someone could have alerted us about this  
ahead of RC2. So it goes.

> I sent the updated patch a month ago, and then the next week Stas  
> asked some questions off-list, for clarification, etc. and then said  
> that the patch "looks good" and "since it appears to fix things I  
> think it can be applied." That's the only feedback I had really, and  
> Dmitry mentioned that it "breaks" about 30 tests (I'd consider them  
> broken now, to match the code, however ;-)), which I knew would have  
> to be updated.  I didn't try to fix them yet, since I didn't know if  
> the changes would finally be applied or not.  I was going to bring  
> it up again but then it was too close to RC2.
>
> There were some e-mails on the subject that I didn't follow up on  
> (nothing major, just comments), including one of yours I think.   
> Anyway, I guess I/we can wonder about RC3 now?  Again, the *very  
> minor* modifications only help to ensure the [usual] long-standing  
> behavior on all platforms -- e.g. most users would see no change  
> from 5.2 or prior.

@Stas/Dmitry: ?

regards,
Lukas Kahwe Smith
mls@pooteeweet.org

---

## #43912 — Stanislav Malyshev — 2009-05-12T18:12:11+00:00

URL: https://externals.io/message/43869#43912 · In reply to: #43898

Hi!

> @Stas/Dmitry: ?

As I said, I'm ok with committing this, provided all the tests are 
fixed, etc.
-- 
Stanislav Malyshev, Zend Software Architect
stas@zend.com   http://www.zend.com/
(408)253-8829   MSN: stas@zend.com

---

## #43921 — Lukas Kahwe Smith — 2009-05-13T09:39:49+00:00

URL: https://externals.io/message/43869#43921 · In reply to: #43912

> Hi!
>
> > […]
>
> As I said, I'm ok with committing this, provided all the tests are  
> fixed, etc.

I do not know what the changes do. AFAIK they revert things to be more  
BC compliant with 5.2 and if that is the case, then there is no reason  
from an RM perspective to stop this change, if you guys feel its the  
right choice.

If tests need fixing and you dont have the time for this, I am sure we  
will find people that do.

regards,
Lukas Kahwe Smith
mls@pooteeweet.org

---

## #43998 — Stanislav Malyshev — 2009-05-15T20:41:41+00:00

URL: https://externals.io/message/43869#43998 · In reply to: #43921

Hi!

> If tests need fixing and you dont have the time for this, I am sure we 
> will find people that do.

I definitely don't have time for doing it, so I'd be happy if somebody 
else does that thing.
-- 
Stanislav Malyshev, Zend Software Architect
stas@zend.com   http://www.zend.com/
(408)253-8829   MSN: stas@zend.com

---

## #44086 — Lukas Kahwe Smith — 2009-05-28T16:28:12+00:00

URL: https://externals.io/message/43869#44086 · In reply to: #43888

> Hi Lukas,
>
> ----- Original Message -----
> From: "Lukas Kahwe Smith"
> Sent: Monday, May 11, 2009
>
> > […]
>
> No, can't say that all is well. :-/  Nothing was changed yet, sorry  
> if you misunderstood.  (BTW, it's not rounding/parsing, but  
> conversion/casting of floats->integers... :-))
>
> I sent the updated patch a month ago, and then the next week Stas  
> asked some questions off-list, for clarification, etc. and then said  
> that the patch "looks good" and "since it appears to fix things I  
> think it can be applied." That's the only feedback I had really, and  
> Dmitry mentioned that it "breaks" about 30 tests (I'd consider them  
> broken now, to match the code, however ;-)), which I knew would have  
> to be updated.  I didn't try to fix them yet, since I didn't know if  
> the changes would finally be applied or not.  I was going to bring  
> it up again but then it was too close to RC2.
>
> There were some e-mails on the subject that I didn't follow up on  
> (nothing major, just comments), including one of yours I think.   
> Anyway, I guess I/we can wonder about RC3 now?  Again, the *very  
> minor* modifications only help to ensure the [usual] long-standing  
> behavior on all platforms -- e.g. most users would see no change  
> from 5.2 or prior.
>
> I'll try to be sure to do what I can to take care of anything now,  
> since I shouldn't be distracted with other stuff like leading up to  
> RC2...


I never saw anything about this issue being fixed with a commit yet.
Could someone post the patch, so the fix can be put into CVS and  
people can take care of updating the relevant tests?

regards,
Lukas Kahwe Smith
mls@pooteeweet.org

---

## #44090 — Matt Wilmas — 2009-05-29T03:05:48+00:00

URL: https://externals.io/message/43869#44090 · In reply to: #44086

Hi Lukas, all,

----- Original Message -----
From: "Lukas Kahwe Smith"
Sent: Thursday, May 28, 2009

>
> 
>
> > […]
>
>
> I never saw anything about this issue being fixed with a commit yet.
> Could someone post the patch, so the fix can be put into CVS and  people 
> can take care of updating the relevant tests?

Ahhh, sorry I hadn't been in touch, but I actually have good news this time. 
:-)  I finished updating most of the tests a few days back, and took care of 
the last couple now.  There were 40-something, though there may be some I 
missed -- only running ~57% of 5.3's tests on Windows with limited 
extensions, etc.  I fixed a few 64-bit tests (hopefully) I noticed along the 
way also.  Well, when the changes get committed, anyone can let me know if 
others are failing, and I'll try to take care of them.  It's generally just 
taking the expected result from 5.2's tests...  I'd think there'd have been 
some sort of red flags with such different results on tests that have been 
added since that 5.3 change. :-)

I just uploaded the updated patch -- I haven't made any changes, just 
against the latest CVS (had a couple conflicts).  Don't have time now, but 
will later send a reply about it to my specific thread about it [1] (sounds 
like it's mostly good to go though).  I can also send a patch with the test 
changes in case anyone wants to take a look.

[1] http://marc.info/?l=php-internals&m=123939280802580&w=2

> regards,
> Lukas Kahwe Smith
> mls@pooteeweet.org

- Matt