| Pike v7.5 release 12 running Hilfe v3.5 (Incremental Pike Frontend) | > Regexp.PCRE rx=Regexp.PCRE( "("([^"\\]|\\.)*")" ); | > rx->split( """ + "x"*0xE800 ); | zsh: segmentation fault pike
Ouch. That lowered my esteem for PCRE quite a bit. It ought to be reported to them.
Yes, if it is a PCRE problem and not a Pike problem... But I can't see how it can be a Pike problem. Does Pike switch method somewhere depending on string length?
/ Mirar
Previous text:
2003-10-27 12:49: Subject: Re: Are strings generally limited?
Mirar mirar-yg@mirar.org wrote:
| Pike v7.5 release 12 running Hilfe v3.5 (Incremental Pike Frontend) | > Regexp.PCRE rx=Regexp.PCRE( "("([^"\\]|\\.)*")" ); | > rx->split( """ + "x"*0xE800 ); | zsh: segmentation fault pike
Hmm?
Ouch. That lowered my esteem for PCRE quite a bit. It ought to be reported to them.
/ Brevbäraren
Well, yes, very small strings are allocated with block-alloc (less than 12 bytes or so, perhaps? I don't remember the exact cutof value)
I very much doubt that that is the problem, though.
/ Per Hedbor ()
Previous text:
2003-10-27 12:55: Subject: Re: Are strings generally limited?
| Pike v7.5 release 12 running Hilfe v3.5 (Incremental Pike Frontend) | > Regexp.PCRE rx=Regexp.PCRE( "("([^"\\]|\\.)*")" ); | > rx->split( """ + "x"*0xE800 ); | zsh: segmentation fault pike
Ouch. That lowered my esteem for PCRE quite a bit. It ought to be reported to them.
Yes, if it is a PCRE problem and not a Pike problem... But I can't see how it can be a Pike problem. Does Pike switch method somewhere depending on string length?
/ Mirar
Well, the first step would be to whip together a plain C program that does the same and verify that it fails too.
/ Martin Stjernholm, Roxen IS
Previous text:
2003-10-27 12:55: Subject: Re: Are strings generally limited?
| Pike v7.5 release 12 running Hilfe v3.5 (Incremental Pike Frontend) | > Regexp.PCRE rx=Regexp.PCRE( "("([^"\\]|\\.)*")" ); | > rx->split( """ + "x"*0xE800 ); | zsh: segmentation fault pike
Ouch. That lowered my esteem for PCRE quite a bit. It ought to be reported to them.
Yes, if it is a PCRE problem and not a Pike problem... But I can't see how it can be a Pike problem. Does Pike switch method somewhere depending on string length?
/ Mirar
I made a new testcase for the PCRE testsuite. It failed.
/ Martin Nilsson (saturator)
Previous text:
2003-10-27 19:38: Subject: Re: Are strings generally limited?
Well, the first step would be to whip together a plain C program that does the same and verify that it fails too.
/ Martin Stjernholm, Roxen IS
Well, if it was the glue, then he would have made the same mistake as you. (I am talking about the PCRE testsuite that comes with PCRE)
/ Martin Nilsson (saturator)
Previous text:
2003-10-28 19:28: Subject: Re: Are strings generally limited?
Is it because PCRE fails or the glue fails?
/ Mirar
Ah, I see. Excellent. :-)
/ Mirar
Previous text:
2003-10-28 19:36: Subject: Re: Are strings generally limited?
Well, if it was the glue, then he would have made the same mistake as you. (I am talking about the PCRE testsuite that comes with PCRE)
/ Martin Nilsson (saturator)
http://pike.ida.liu.se/generated/testinput6.gz if you are interested. I sent a copy to ph10@cam.ac.uk.
/ Martin Nilsson (saturator)
Previous text:
2003-10-28 19:37: Subject: Re: Are strings generally limited?
Ah, I see. Excellent. :-)
/ Mirar
pike-devel@lists.lysator.liu.se
-
Martin Nilsson (saturator) ＠ Pike (-) developers forum -
Martin Stjernholm, Roxen IS ＠ Pike developers forum
-
Mirar ＠ Pike developers forum
-
Per Hedbor () ＠ Pike (-) developers forum