Re: Re: OCI8 1.1 announce

Hi Manuel, Thanks for the feedback. Does anyone here, who knows configure better than me, have some insight into the host_os mismatch? Is this with all extensions? Andi At 12:36 PM 9/9/2005, Manuel Lemos wrote:

Hello,

Yesterday OCI8 extension have been updated in the PHP CVS (HEAD only). This updated driver resolves a large amount of bug reports and adds some improvements both functionality and performance, and much better documentation (see here: http://php.net/oci8). A lot of time has been invested in updating this driver including help from Wez Furlong and the OCI team which gave a lot of feedback on how to do things the best and most efficient way.

I have just tried this version and at least one of two problems that I had with version 1.0 was solved. The problem that seems solved used to crash PHP if ORACLE_HOME and ORACLE_SID were not set before running PHP. I have tested it with Metabase OCI driver class and previously have documented the problem in the OCI driver notes of Metabase documentation. http://www.meta-language.net/metabase.html#232.2.4 Metabase uses PHP putenv() function to set these variables but with OCI8 1.0 extension not setting these variables before running PHP was causing segmentation or the error ORA-03117: two-task save area overflow. Anyway, now that no longer seems to be a problem, as in past versions prior to PECL OCI8 1.0 extension. The other problem that I am still having is about host_os variable in the configure script being set to linux but the configure script generated by phpize expects linux-gnu. The consequence of this is that even when you use --enable-shared as configure argument makes it always build as static. I use SuSE Linux Professional 9.0 with PHP 4.3.11-1 RPM archives provided in SuSE FTP site here: ftp://ftp.suse.com/pub/projects/apache/mod_php4/9.0-i386/ To solve the problem, I had to hack the configure script generated by phpize adding a few case options to consider linux as host_os alias of linux-gnu . Using that hack the extension builds and installs properly, passing 15 out of 16 tests of Metabase driver class test suite. The only test that fails is due to a known "feature" of Oracle of storing NULL when you mean to store an empty string on a VARCHAR fields. So, all works well as expected. Congratulations! ;-) -- Regards, Manuel Lemos PHP Classes - Free ready to use OOP components written in PHP http://www.phpclasses.org/ PHP Reviews - Reviews of PHP books and other products http://www.phpclasses.org/reviews/ Metastorage - Data object relational mapping layer generator http://www.meta-language.net/metastorage.html -- PECL development discussion Mailing List (http://pecl.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php