On Mon, Mar 3, 2008 at 10:38 AM, Chris Chabot (JIRA) <[EMAIL PROTECTED]> wrote:
> > I'll fix up the copied-features issue and some other little things and > re-submit a patch. That doesn't mean other things wont be picked up and in > fact there's bound to be a few small bugs hiding in it too, but since a good > few people wanted to see this included sooner rather then later, i won't > block submitting it on having everything completely done. If we get it good > enough for initial inclusion other people can start looking at it and maybe > contribute where possible too. Like I said -- the other stuff is just suggestions. The copied features issue is the only one I'd consider holding off applying the patch for. W.I.P. (work in progress), i also felt that some re-factoring would be a > good thing to do next, however as a few people suggested on the mailing > list, it's better to submit a patch now and get the ball rolling then to > wait for it to be completely Done(Tm). Agreed. > Since when i began this i had no idea how the java server worked, i just > jumped in somewhere in the middle and trusted in the fact that by reading > the java base line per line a bigger picture of what it did exactly would > appear; Fortunately that worked, so i'm in a lot better position now to > start making bigger changes to it; (Zen mode) To deviate from the way you > must first know the way right? :) > > Documentation is in the same line something i wanted to leave until i > would have a much clearer picture of what was what, else it would've become > a byte filler and not something sensible Fair enough :) > - define() is horribly slow for no good reason in PHP > > - include_once is also horribly slow for no good reason > > Actually to my best knowledge the performance impact of include_once was > fixed somewhere in the php 5.1 era, however i do realize as i'm writing > this, that some people might still be stuck there, so i'll fix it up for > that case. Unfortunately, this isn't the case when you use APC or Zend Cache, because it prevents several important optimizations from taking place. Anyway, it's a performance issue and as such should probably be ignored until it actually proves to be a problem. > - Have you considered using __autoload > > > I truly love the new magic functions in php 5, however i've found in the > past that the performance of __autoload is completely horrific, i'll see how > it behaves now with php 5.2, if it improved any it is a good option, > otherwise if its still as bad as it was, i'd rather not :) I haven't used PHP since the early 5.2 release (about a year ago), and then autoload behavior seemed pretty reasonable for a medium sized site. When I was at yahoo we generally avoided it, but there were some improvements made recently. It's worth an experiment, I suppose, but I wouldn't change all my code just for it :)
