We currently run PHP 5.2.17 from atomic + several libs (mcrypt, dom, mysqli, etc).

I'm assuming the 5.3 RPMs are in "testing" because they are not ready for a production environment, is this correct?

If so, is there a known set of problems (e.g: segfaults) with the 5.3 RPMS ... or are these far enough along in testing that upgrading is reasonably safe?

I'm hoping it's just a housekeeping problem and that they're ready to be moved to the main repository whenever someone has time.

thanks

That is a great question, one I was just about to bring up for public comment. Its currently in there because public comment on standardizing the repo on 5.3 hasnt been really started yet.

So lets open it up for the final approval. Once this is done, the PHP 5.2 branch will officially be retired and no longer maintained. Any issues, please bring them up here and lets give the review til the end of the month.

I've had 5.3.3 installed on my production servers for some time. Works great. I can't go to 5.3.5 because there's no eaccelerator package for it.

Great, so if there are no other objections PHP 5.2 will be retired sometime this week.

Only question i have is will this effect zend and ioncube by going to the 5.3 branch? i thought i saw a post some where that mentioned issues with them on 5.3 but was'nt sure and just looked around to find it but had no luck.

i have a couple customers that have to use zend and a couple that are using ioncube and last thing i need to do is break their stuff by going to 5.3.

Just that is no / never will be a zend-optimizer for 5.3. THey have a different thing (requires key) that we arent allowed to distribute.

Zend Optimizer is part of Zend Server now. AFAIK this is the only way to run it under 5.3
http://www.zend.com/en/products/server/#Performance
http://www.zend.com/en/products/server-ce/downloads

I am surprised they won't allow distros of the Community Edition (not that I would use it). There is a Zend yum repo if you really want Zend Server CE.
http://files-source.zend.com/help/Zend- ... lation.htm

Just bringing up the issue of getting php-fpm to work on the ART 5.3 distro.

I personally had to switch to using the REMI Repo to get 5.3.4+ with fpm-php and will like to suggest that if pushing 5.3 into the mainstream, then the build instruction for fpm which are already in the spec file but not working should be review and fixed if possible.

I suppose an alternative is to remove them altogether.

Cheers.

Thats not so much of a barrier on moving from 5.2 to 5.3. Lets not hijack the thread here.

I don't see it as hijacking the thread as I believe you asked for input on issues and I raised one of components in the build spec not working.

Now, it may not be a much of a barrier as you say although if the items are in the spec file as they are, one would expect someone must have put them there presumably with the intention that they would be built and that the item would work.

Anyway, it has been raised and evaluated by you and there can't be any harm in that ... it may be that it reminds you to fix it at some point.

So thats +1 against going with 5.3 then?

+1 for including PHP 5.3. We are using the atomic-testing packages for some months now without any problems.

A suggestion: create an Xcache package. It's getting increasingly popular as a faster/lightweight PHP opcode cache

I don't think so. Afterall, you said it was not so much of a barrier and I agree.

I just raised it within the context of me thinking the testing repo was there for people like me to try what is there and give feedback on what works and what doesn't. I happened to notice that FPM was not getting built even though the instructions to build were present and thought I should bring it to your attention. I now assume you will review this and decide on how to proceed.

It isn't a major thing for me as I have found an alternative but I would of course prefer to only use one repo if possible.
So if convenient and worth fixing, great. If not, life goes on and it can be sorted out later if deemed worth it.

ok so +2 for 5.3 then?

+1 for going to 5.3

That's +3 in my book, but who's counting?