[rkward-devel] php (5.2.11) backend error with R 2.10 on KDE3

Prasenjit Kapat kapatp at gmail.com
Thu Dec 3 20:38:17 UTC 2009


Hi,

Sorry it took me awhile to get back:

On Thu, Nov 26, 2009 at 5:38 AM, Thomas Friedrichsmeier
<thomas.friedrichsmeier at ruhr-uni-bochum.de> wrote:
> Hi,
>
> On Wednesday 25 November 2009, Prasenjit Kapat wrote:
>> The PHP-backend has died unexpectedly. The current output buffer is shown
>>  below: STDIN is at eof
>
> that's bad. Could you try the following patch? It's just a guess, but perhaps
> this does the trick.
>
[snip]
>
> +usleep (500000);
> +
[snip]
>
> Do you have any other version of PHP available to test?

OK, I'll break down into these categories:
{w/,w/o}us: {with, without} the above usleep() patch
p-x.y.z:        with php version x.y.z


w/o us + p-5.1.6 (from CentOS 5.4) : works.
w/o us + p-5.2.10 (for RHEL5 'remi') : works. [*]
w/o us + p-5.2.11 (/usr/local/bin/php) : sometime does sometime doesn't.
w/o us + p-5.3.0-2 (for RHEL5 'remi') : works [*]
w/o us + p-5.3.1-1 (for RHEL5 'remi') : does not work, segfaults. [*,+]

(a complete make distclean was done before compiling w/us - may not
have been needed)

w/us + p-5.1.6 (from CentOS 5.4) : works.
w/us + p-5.2.10 (for RHEL5 'remi') : works. [*]
w/us + p-5.2.11 (/usr/local/bin/php) : sometime does sometime doesn't.
w/us + p-5.3.0-2 (for RHEL5 'remi') : works [*]
w/us + p-5.3.1-1 (for RHEL5 'remi') : does not work, segfaults. [*,-]

The usleep line does not seem to have any effect. I didn't collect
full debug outputs.

Our dept will update to 2.10.1 soon, so for the time being (or may be
even for long) I'll stick with one of the working phps, we can debug
later.

Regards,
-- 
Prasenjit

[*] I am not sure what this 'remi' thing is, anyway, the php-cli
(x86_64 versions) packages used were obtained from here (searching for
php-cli and checking only RHEL5 and CentOS 5):
http://rpm.pbone.net/index.php3/stat/2/simple/2

[+] segfaults with:
output without receiver'
 *** caught segfault ***
address 0x75cb40, cause 'invalid permissions'

[-] segfaults with:
output without receiver'
 *** caught segfault ***
address 0x20000007365, cause 'memory not mapped'




More information about the Rkward-devel mailing list