The fix looks to be safe enough to MFH, I see no API breakage.
On 5-Jan-09, at 3:36 PM, Felipe Pena wrote:
Hi,
Em Sáb, 2009-01-03 às 11:20 -0800, shire escreveu:
Felipe Pena wrote:
Em Sáb, 2009-01-03 às 10:31 -0200, Felipe Pena escreveu:
Em Sáb, 2009-01-03 às 10:27 -0200, Felipe Pena escreveu:
Em Sex, 2009-01-02 às 21:03 -0800, shire escreveu:
I've created a patch for bug #46701 (http://bugs.php.net/bug.php?id=46701
) but it requires Zend changes, it can be found at the links
below for all branches. I've verified all tests pass. We may
want to verify it for other architectures due to the nature of
it being a float conversion problem, I tested on Intel OS X
10.5.5. But this should at least be equivalent to
$array[intval($double)] now.
http://tekrat.com/patches/bug46701.php6.patch
http://tekrat.com/patches/bug46701.php53.patch
http://tekrat.com/patches/bug46701.php52.patch
I suppose that should we also change the fetch (ZEND_FETCH_DIM_*)?
And probably:
case IS_DOUBLE:
index = (long)Z_DVAL_P(dim);
goto num_index;
And zend_execute.c (zend_fetch_dimension_address_inner) will be
superfluous with this change.
Errr, I mean, the case IS_DOUBLE stuff in
zend_fetch_dimension_address_inner will be superfluous.
In the end, I see that it requires the macro too, hehe.
So, a little change in your patch:
- case IS_DOUBLE:
- index = (long)Z_DVAL_P(dim);
+ case IS_DOUBLE: {
+ DVAL_TO_LVAL(Z_DVAL_P(dim), index);
goto num_index;
-
+ }
Hey, Good point Felipe, thanks! I've updated my patches as well
to include this change and verified all the tests.
I've committed it in 5_3 and HEAD. Ilia, should it be in 5_2 too?
Thanks for the patch, Shire. ;)
--
Regards,
Felipe Pena
Ilia Alshanetsky
--
PHP Internals - PHP Runtime Development Mailing List
To unsubscribe, visit: http://www.php.net/unsub.php
