Does it pay and time to upgrade php version?

Started by chromerep, December 22, 2012, 04:51:50 AM

Previous topic - Next topic

chromerep

Does it pay and time to upgrade from phph5.3.20 to php5.4.10?Is there any benefit to do?
I use mod_fcgid to run php scripts.

Gregg

I was sort of leaving this one for Mario to answer since he breaths PHP but I guess he's got no opinion either way.

I will say I did but for no other reason than to be using the latest greatest. As of .10 TS I can even run as module in 2.4 which has a huge speed advantage for me, but I do not have a server heavily laden with PHP nor a high traffic volume.

I'm unsure as to whether anything in 5.3 has vanished from 5.4 so not sure if it could break some script that works perfectly good in 5.3. Best I can say is install it side-by-side, change the pointers in your mod_fcgid config, Restart Apace and give it a test. You can always change the pointers back to 5.3 and restart should you find problems.

mario

PHP 5.3 will be discontinued soon (End of Lifetime). For your scripts there won't be a change. All my scripts that run smooth on 5.3 run very well with 5.4
Well some functions do not exist anylonger / see the E_DEPRECATED errors in 5.3. If you have a lot of them you have to fix them first.
You should upgrade since there will be PHP 5.5 and than there won't be any support on 5.3

chromerep

I think there is no choice to upgrade to php5.4.x since php5.3.x will be at the EOL.
I use Vhost to run some php scripts that can't run well at php5.4.x.There is no idea to solve php scripts issues.I am poor at php scripting.

And,I should take a look at the running status between phph 5.4.x and apache 2.4.3.

If I run mod_fcgid via php-cgi.exe,to upgrade to php5.4.x will be faster or not?

mario

PHP Devs say that it is faster on windows than 5.3. My personal benchmark showed a little improvement.

SInce you run it over fcgid, you could run 5.3 and 5.4 parallel for testing. I used to set up a second vhost with the newer PHP version, but the same document root.