Hi!
Bug #48247 appears to be not entirely fixed still, it still produces a
torrent of warnings in certain configs (namely, take php.ini-production
and edit it to enable error_log, don't touch anything else - I get 400K
worth of warnings). Attached patch should fix it. Please tell if there's
some objection to it, otherwise I'll commit it tomorrow.
Stanislav Malyshev, Zend Software Architect
stas@zend.com http://www.zend.com/
(408)253-8829 MSN: stas@zend.com
Hi!
Bug #48247 appears to be not entirely fixed still, it still produces
a torrent of warnings in certain configs (namely, take php.ini-
production and edit it to enable error_log, don't touch anything
else - I get 400K worth of warnings). Attached patch should fix it.
Please tell if there's some objection to it, otherwise I'll commit
it tomorrow.
So whats going on here?
I noticed that the bug report hasn't been reopened .. but I think
there was also not commit yet .. ?
regards,
Lukas Kahwe Smith
mls@pooteeweet.org
Hi!
So whats going on here?
I noticed that the bug report hasn't been reopened .. but I think there
was also not commit yet .. ?
Well, it's not exactly the same bug - crash is not there anymore, but
infinite loop is. There's bug http://bugs.php.net/bug.php?id=48534 about
it so I decided not to reopen this one.
Stanislav Malyshev, Zend Software Architect
stas@zend.com http://www.zend.com/
(408)253-8829 MSN: stas@zend.com
Bug #48247 appears to be not entirely fixed still, it still produces a torrent
of warnings in certain configs (namely, take php.ini-production and edit it to
enable error_log, don't touch anything else - I get 400K worth of warnings).
Attached patch should fix it. Please tell if there's some objection to it,
otherwise I'll commit it tomorrow.
The attached patch does fix one of the symtomps of the issue, but not
the real problem. The real problem here is that from an error handler,
another error is thrown. Instead of fixing this as a hack in any
extension that has influence on formatting of error messages or error
handling (through overriding the callback), a fix should be done in the
error handling mechanism in order to prevent an error message generating
another error message.
regards,
Derick
--
http://derickrethans.nl | http://ezcomponents.org | http://xdebug.org
twitter: @derickr
Bug #48247 appears to be not entirely fixed still, it still
produces a torrent
of warnings in certain configs (namely, take php.ini-production and
edit it to
enable error_log, don't touch anything else - I get 400K worth of
warnings).
Attached patch should fix it. Please tell if there's some objection
to it,
otherwise I'll commit it tomorrow.The attached patch does fix one of the symtomps of the issue, but not
the real problem. The real problem here is that from an error handler,
another error is thrown. Instead of fixing this as a hack in any
extension that has influence on formatting of error messages or error
handling (through overriding the callback), a fix should be done in
the
error handling mechanism in order to prevent an error message
generating
another error message.
So lets delay fixing this issue until 5.3.1? Of course it can be fixed
in HEAD right now :)
regards,
Lukas Kahwe Smith
mls@pooteeweet.org
Hi!
So lets delay fixing this issue until 5.3.1? Of course it can be fixed
in HEAD right now :)
Delaying the fix would mean we release 5.3.0 that in mostly default
config has an infinite loop bug. Maybe we can at least have temp fix and
then when we have a proper one then remove it (or leave it in, it
wouldn't hurt anybody)?
For 5.2 it's even worse since I understand 5.2 is not intended to be
developed past 5.2.10, except for security fixes.
Stanislav Malyshev, Zend Software Architect
stas@zend.com http://www.zend.com/
(408)253-8829 MSN: stas@zend.com
hi,
Hi!
So lets delay fixing this issue until 5.3.1? Of course it can be fixed in
HEAD right now :)Delaying the fix would mean we release 5.3.0 that in mostly default config
has an infinite loop bug. Maybe we can at least have temp fix and then when
we have a proper one then remove it (or leave it in, it wouldn't hurt
anybody)?
I already got complains about this infinite loops (there is a bug report too).
For 5.2 it's even worse since I understand 5.2 is not intended to be
developed past 5.2.10, except for security fixes.
5.2 is likely to be released on Thursday, please, try to figure out a
solution before that or to convince Ilia to wait, thanks :)
Cheers,
Pierre
Hi!
5.2 is likely to be released on Thursday, please, try to figure out a
solution before that or to convince Ilia to wait, thanks :)
Well, there is a patch I sent, which fixes the immediate problem, if
Ilia OKs it I can commit it any time. The comprehensive fix would take
some time though.
Stanislav Malyshev, Zend Software Architect
stas@zend.com http://www.zend.com/
(408)253-8829 MSN: stas@zend.com
Hi!
So lets delay fixing this issue until 5.3.1? Of course it can be
fixed in HEAD right now :)Delaying the fix would mean we release 5.3.0 that in mostly default
config has an infinite loop bug. Maybe we can at least have temp fix
and then when we have a proper one then remove it (or leave it in,
it wouldn't hurt anybody)?
For 5.2 it's even worse since I understand 5.2 is not intended to be
developed past 5.2.10, except for security fixes.
so if this issue is also in 5.2 .. i feel even less inclined to delay
a stable 5.3 release because of this issue. have we made it worse in
our attempts to fix it? the fact that this bug isnt fixed yet could
mean several things like:
- the issue affects few users
- the issue is so complex that we continuously try to fix it and fail
both issues do not sound like good reason to delay 5.3 from where i
stand.
regards,
Lukas Kahwe Smith
mls@pooteeweet.org
Hi!
so if this issue is also in 5.2 .. i feel even less inclined to delay a
stable 5.3 release because of this issue. have we made it worse in our
attempts to fix it? the fact that this bug isnt fixed yet could mean
No, it was crash before, now it doesn't crash for me but produces 400K
of errors. One can argue it's better now, though pretty marginally as in
both cases system can not be considered stable. Of course, there's an
easy workaround of setting the timezone.
- the issue affects few users
- the issue is so complex that we continuously try to fix it and fail
We didn't "continuously" try to fix it - we applied one fix for the
crash (which was caused by different reason than the loop), and need
another for the loop.
As for why people don't complain more - I guess it's because it has
obvious workaround and once people don't have it anymore many are too
lazy to complain about the problem which now is not their problem :)
both issues do not sound like good reason to delay 5.3 from where i stand.
We don't need to delay it if we decide to commit my (incomplete) fix
now. Anyway, it's your decision, one of the three: 1. leave it as is, 2.
commit incomplete fix now, 3. wait for complete fix. I agree 3 is the
worst, I'd go with 2 but if you decide to go with 1 that's OK.
Stanislav Malyshev, Zend Software Architect
stas@zend.com http://www.zend.com/
(408)253-8829 MSN: stas@zend.com
We don't need to delay it if we decide to commit my (incomplete) fix
now. Anyway, it's your decision, one of the three: 1. leave it as
is, 2. commit incomplete fix now, 3. wait for complete fix. I agree
3 is the worst, I'd go with 2 but if you decide to go with 1 that's
OK.
Just to clarify how I see the job of an RM. Its not chief decision
maker. Its decision facilitator. Only when with reasonable amount of
time and input from the community no agreement/compromise can be
found, I can see the RM act as a tie breaker. As such I prefer to get
feedback from as many people as possible in a clear and concise manner
that makes it easy for us RM's to understand the view points of the
community, so that we can then simply sum of the conclusion of the
discussion and ask someone to make the necessary changes.
regards,
Lukas Kahwe Smith
mls@pooteeweet.org
Hi!
so if this issue is also in 5.2 .. i feel even less inclined to
delay a stable 5.3 release because of this issue. have we made it
worse in our attempts to fix it? the fact that this bug isnt fixed
yet could meanNo, it was crash before, now it doesn't crash for me but produces
400K of errors. One can argue it's better now, though pretty
marginally as in both cases system can not be considered stable. Of
course, there's an easy workaround of setting the timezone.
It seems only you and Derick have expressed an opinion.
If no one else speaks up I would tend to leave as is and fix it in
5.3.1 (and HEAD asap) and simply make sure that we put it in bold
letters in the UPGRADING guide that you really should set the timezone.
regards,
Lukas Kahwe Smith
mls@pooteeweet.org
It seems only you and Derick have expressed an opinion.
If no one else speaks up I would tend to leave as is and fix it in 5.3.1
(and HEAD asap) and simply make sure that we put it in bold letters in the
UPGRADING guide that you really should set the timezone.
btw, I wished we could raise a fatal error in this case, ok, it is too
late but I can have wishes :)
--
Pierre