Litespeed and PHP 4.4.7 end of life
- Thread starter brrr
- Start date
Aha. Thanks for the info.
But how much faster is 4.4.7 over PHP 5.2.4? 10%? or >50%?
If the performance margin isn't substantial, I'd suggest that the potential benefits of moving to 'native' support for 5.2.4 now would outweigh any potential performance hit.
The benefits include less hassle for LSWS customers, most of which (judging from forum posts) seem to end up compiling PHP 5 support in anyhow, which sometimes gets messy.
Some mainstream PHP apps (like MediaWiki) that potential LSWS users might want to get into also now won't run on PHP4, unless you revert to older versions of the software.
And given the 4.4.7 end of life in Dec 2007, by years end I suspect most folks won't want to be running un-supported software like PHP4 in future, esp business users.
So maybe the time to transition LSWS to PHP 5 is now?
But how much faster is 4.4.7 over PHP 5.2.4? 10%? or >50%?
If the performance margin isn't substantial, I'd suggest that the potential benefits of moving to 'native' support for 5.2.4 now would outweigh any potential performance hit.
The benefits include less hassle for LSWS customers, most of which (judging from forum posts) seem to end up compiling PHP 5 support in anyhow, which sometimes gets messy.
Some mainstream PHP apps (like MediaWiki) that potential LSWS users might want to get into also now won't run on PHP4, unless you revert to older versions of the software.
And given the 4.4.7 end of life in Dec 2007, by years end I suspect most folks won't want to be running un-supported software like PHP4 in future, esp business users.
So maybe the time to transition LSWS to PHP 5 is now?
Aha. Thanks for the info.
But how much faster is 4.4.7 over PHP 5.2.4? 10%? or >50%?
If the performance margin isn't substantial, I'd suggest that the potential benefits of moving to 'native' support for 5.2.4 now would outweigh any potential performance hit.
The benefits include less hassle for LSWS customers, most of which (judging from forum posts) seem to end up compiling PHP 5 support in anyhow, which sometimes gets messy.
Some mainstream PHP apps (like MediaWiki) that potential LSWS users might want to get into also now won't run on PHP4, unless you revert to older versions of the software.
And given the 4.4.7 end of life in Dec 2007, by years end I suspect most folks won't want to be running un-supported software like PHP4 in future, esp business users.
So maybe the time to transition LSWS to PHP 5 is now?
But how much faster is 4.4.7 over PHP 5.2.4? 10%? or >50%?
If the performance margin isn't substantial, I'd suggest that the potential benefits of moving to 'native' support for 5.2.4 now would outweigh any potential performance hit.
The benefits include less hassle for LSWS customers, most of which (judging from forum posts) seem to end up compiling PHP 5 support in anyhow, which sometimes gets messy.
Some mainstream PHP apps (like MediaWiki) that potential LSWS users might want to get into also now won't run on PHP4, unless you revert to older versions of the software.
And given the 4.4.7 end of life in Dec 2007, by years end I suspect most folks won't want to be running un-supported software like PHP4 in future, esp business users.
So maybe the time to transition LSWS to PHP 5 is now?
I know we always do a compile and never use the default binary, unless under special circumstances.
Yes, I bit the bullet and have started down the ./configure path, first on my CentOS5 dev box, and then if all works OK, onto my CentOS 4.5 prod machine.
But so far - UGH - what a hassle. Multiple ./configure halts - even though I was basing ./configure on the existing fully functional options for another PHP5 install on the same CentOS5 box, first the PHP ./configure run wanted pcre-devel, then aspell-devel, then it barfed again on something else having the wrong base path. Lots of googling and yumming and rpm finding later (=about 1 or 2 hours of scratching around) I finally got a PHP 5.2.4 binary to miraculously emerge from the build process on my dev server.
Now I have to test out whether the new PHP binary will stably work with LSWS and eAccelerator as-is or I will need to recompile eAccel too for this PHP5 instance. If all this works out, then I'll move to attempt the recompile process all over again on the CentOS 4.5 production site.
Compare this to the about **2 minutes** it took to originally install LSWS w PHP 4.4.7 and eAccelerator on both of my servers and you see why I would love for PHP5 to be part of the LSWS default install.
But so far - UGH - what a hassle. Multiple ./configure halts - even though I was basing ./configure on the existing fully functional options for another PHP5 install on the same CentOS5 box, first the PHP ./configure run wanted pcre-devel, then aspell-devel, then it barfed again on something else having the wrong base path. Lots of googling and yumming and rpm finding later (=about 1 or 2 hours of scratching around) I finally got a PHP 5.2.4 binary to miraculously emerge from the build process on my dev server.
Now I have to test out whether the new PHP binary will stably work with LSWS and eAccelerator as-is or I will need to recompile eAccel too for this PHP5 instance. If all this works out, then I'll move to attempt the recompile process all over again on the CentOS 4.5 production site.
Compare this to the about **2 minutes** it took to originally install LSWS w PHP 4.4.7 and eAccelerator on both of my servers and you see why I would love for PHP5 to be part of the LSWS default install.
I eventually got a functional binary built on CentOS 5 after going back to the original PHP and LS SAPI sources, rather than the medialayer pre-assembled lsphp kit, which (for me at least) did not cut the mustard
I guess you could take the same approach to building PHP5 into a LSWS installer as you have with 4.4.7 - not all modules are in the 4.4.7 installer but you made a judicious selection with 4.4.7 that delivers enough functionality for most users to get straight down to business with LSWS with their standard PHP apps (forums, CMS's, wiki's, blogging platforms etc) without hours of error-prone futzing about trying to roll-their-own PHP
I guess you could take the same approach to building PHP5 into a LSWS installer as you have with 4.4.7 - not all modules are in the 4.4.7 installer but you made a judicious selection with 4.4.7 that delivers enough functionality for most users to get straight down to business with LSWS with their standard PHP apps (forums, CMS's, wiki's, blogging platforms etc) without hours of error-prone futzing about trying to roll-their-own PHP