Re: Zend Framework...where to start? -- don't.

Core files are what my plans include too. Bastien Sent from my iPod On Jan 14, 2009, at 9:26 PM, "Kyle Terry" <kyle@kyleterry.com> wrote:

On Wed, Jan 14, 2009 at 6:17 PM, Paul M Foster <paulf@quillandmouse.comwrote:

On Wed, Jan 14, 2009 at 01:39:02PM -0800, Daevid Vincent wrote:

Not to start a Holy War (as these "to framework" or "not to framework" debates often turn into), but I personally had a horrible experience

with

using frameworks. I was forced to use Symfony at my last job and it

was so

cumbersome and slow to do even the simplest things. The whole MVC

thing

can be overkill. Plus the learning curve can be quite steep. Then if

you

want to hire other developers to work with you, you have to train them

and

let them ramp up on not only the framework but also your core project

too!

More wasted time. The pages are significantly slower than straight PHP by orders of magnitude: [1]http://paul-m-jones.com/?p=315

What a great link! I've never seen this kind of comparison before. HTML is 70% faster than straight PHP, and the frameworks (even codeigniter) deliver less than 20% of the performance of straight PHP.

The basic problem with frameworks is they try to be one thing for all people. This carries a lot of baggage with it. There's a lot of crap

you

end up pulling in that you don't want/need. Plus if you want to

deviate at

all, you either have to roll your own, or sometimes you simply just

can't.

They seem attractive with all their plugins and stuff, but honestly, rarely do the plugins do EXACTLY what you want, the way you want. It

might

be as simple as trying to change the look/feel of a button or

something

and you'll find out that you can't -- so now you have this website

that

has this section that doesn't look like the rest of your site. And if

you

find a bug, you have to try to either fix it yourself and then keep

those

changes migrated into new updates, or submit it to the developer and

hope

they implement them (and trust me, you can submit to them and have

them

rejected for all sorts of lame reasons -- even though the work has

been

done and you're using it!) I advise against it. Just follow good practices and use thin wrappers

and

functions. Don't get all OO googlie eyed and try to over-engineer and over-OO the code. OO is great for some things (like a User class) but don't start making some OO page renderer or form builder. Don't fall

into

the DB Abstraction trap either -- just use a wrapper around your DB

calls

(see attached), so you can swap out that wrapper if (and you almost

never

do) you change the DB. Don't be suckered by something like QuickForms

--

you WILL run into limitations that you can't get around and are at

their

mercy. Don't buy the hype that DIV's are the magic bullet and TABLEs

are

"poor design" -- Tables are still the best and most ubiquitous way to align things in a browser agnostic way (including mobile phones, etc.)

and

to layout forms.

I agree and disagree. I agree there's waaay too much herd mentality in the programming field. (Fortunately, Linus Torvalds didn't listen to the academics who insisted that microkernels where THE WAY, or we wouldn't have Linux today.) OO is nifty for some things, but it's certainly not the "fountain of reusability" it was originally promoted to be. And I also agree about tables versus CSS. I can render a page very precisely with tables that would take me hours to get right with CSS. And I really don't give a crap about what "experts" say about anything. I find "experts" to be wrong much of the time. OTOH, I just finished writing about 80K lines of PHP/HTML, all by hand, no OO, no classes, no nothing. Each page in one file, except for a few helper functions in a couple of common files. I wouldn't want to go through that again. I've opted for a framework on rewriting this code, just to cut down on the number of lines of code I have to manually write. But I built my own framework, which doesn't call in 20 files for each page load. Very compact. Probably not suitable for every kind of project, but it works for this. Incidentally, I would differ from the reviewer in the link above only in this respect: He maintains that every line of code adds time. While this is true, I believe it's the number of files which have to be opened which drags down framework numbers the most. When I wrote C code, the CPU would blaze through the actual code, but file opens and reads consumed far more time than in-memory code execution. Paul -- Paul M. Foster -- PHP General Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php

I agree heavily on the file opening part. I hate having to look through a stack trace of 20 or 30 just to track down why an exception was thrown. We are working on moving our entire framework into less files and more of a core set of files that handles a lot of tasks. --Kyle Terry | www.kyleterry.com