Do you ever get halfway through developing...

Soldato
Joined
12 Jan 2004
Posts
3,172
Location
Brighton
And start thinking... what the hell are you doing...

I have spent the last month or two developing a huge PHP Web applications development framework for the hell of it, and I'm kinda wondering what the point of it all is.

I have an idea for a business that I was writing the original web application for, but at a certain point, I thought "I can't be arsed to write all this code by hand" so I started to develop a framework to generate the work itself.

Now I am even deeper into this (my framework already develops models, data access objects (complete with sql database management), model view objects, and now I am automating simple application objects (basically fancy "actions" that do standard things like allow full html interaction with the objects, create read update delete) and I'm suddenly thinking of that shed in "a beautiful mind" full of the insane drawings on the walls...

I just wish I had someone to talk to about this project... /sigh
 
Last edited:
Used to when I was starting out, but now I only develop things if they will serve a purpose or resolve a problem.
 
I realised that a lot of what I am doing is similar to what cakephp offers :S

It'll probably be faster and better engineered than cake though - most things are.

My company has its own proprietary framework because none of the publicly available ones do what we need, in the way we need it done - your own will probably fill that niche for you too.
 
I realised that a lot of what I am doing is similar to what cakephp offers :S
cakephp is pap, it (and the author(s)) like to think it is a suitable framework for "MVC" but in reality it's anything but. Far to much controller and domain logic mish mashed all over the place, and it tries to do far too much for you.

The only frameworks that work are those that are harvested by yourself or (to a lesser degree) forced upon you and everyone else, such as in proprietary languages.
 
Well, at the moment it consists of a series of "generator" programs that take individual XML files and "generate" the php code that is required to produce a certain pattern.

So far I have implemented :

Model (ie, simple class)
SQLDAO (encapsulates model, adds an id, and adds simple shallow sql calls to allow the object itself (but not referenced objects) to serialise itself)
ModelView (which are simple encapsulation classes that produce HTML table headers, table fields and editable fields for the model)
SA (Simple applications, basically just Show/ShowEditable/EditandShow applications that automatically allow an object to be edited from a client-server relationship

I'm planning :
Complex object serialising (deep object serialisation)
Semantic limitations (attach to model, ie age >= 0, text must fit pattern ddmmyy etc)
Users
Workflow
Unique embedded business logic

Not sure what else yet
 
Just make what is needed. No point speculating as to what might be needed in the future, as you'll just end up following an endless trail of niceties.
 
Best way to develop a framework is to actually start off developing an app, once that's done see if you can generalise stuff and extract it from there.

That's how they made RoR (not saying RoR is a great framework, just using it as an example).
 
Ah, but I have an application that it is supposed to produce, and it is producing it, sloooooowly :p

I just lose patience sometimes because it is harder to make the framework than to make the application!
 
Back
Top Bottom