SitePoint’s Advanced PHP Forum is the place to learn best practices in PHP. My handle is Selkirk there.

I am using SimpleTest.

It’s not a question of talent. There is only so much s*t you can handle before you just throw up Some code are like that. Sure one can spend time in trying to fix it. But then when you are fixing something which is poorly architected and designed in the first place, you are most likely going for band-aid solutions.

BTW: This is the reason any programmer worth his salt never says “rewrite” to management. He says “refactor”. That word is very management friendly. How’s that for a paradigm shift

> But software projects are about delivering value to your customers, not about programming.

Yes. Also at the end of the day a programmer needs some job satisfaction out of it. There are human beings on both side of the equation. I think XP achieves both with a remarkable degree – customer satisfaction and programmer satisfaction.

> “infrequently changed data”
I agree. That’s the key point. Not when you are also modifying the file at every request however. Then a DB is a better
choice.

> Just that it is incorrect to think that it can’t scale.

I never even thought that. Definitely you can scale it across multiple machines. My point was that it will have lower performance in hardware utilization and certain prevalent (yet bad) practices makes it harder to scale. The key issue about scalability is how linear is the curve and how long does it maintain the linearity.

BTW: Does PHP have some coding standards or best practices guidelines? That might help in reducing some bad php coding practices I have seen in rather famous Open Source projects.

> I might know more Java than you think.

I wasn’t trying to undermine your Java knowledge.
It’s just a vast subject. It is not possible for everyone to know everything. I am working in Java since ‘96. I don’t claim to know everything, far from it. I am learning everyday.

> I deploy my unit tested PHP application weekly, thank you.

That sounds great. I would love to know what kind of challenges, if any, you face implementing XP practices in PHP.

Are you using phpunit?

Opening and writing to multiple files in the course of serving a single request doesn’t improve scalability. Specially when you are trying to scale across multiple machines.

Coupled with a shared file system, such as NFS, files can be a very appropriate way to share infrequently changed data across several machines.

No amount of imaginary paradigm shift isn’t going to change that. Such practices also makes php a low throughput system.

I’m not trying to make the point that PHP scales better than java. Just that it is incorrect to think that it can’t scale. I am trying to make the point that one uses different techniques in PHP than Java to accomplish the same thing. That is what I meant by paradigm shift.

You have mis-named it. It is called *architecture and design*.
I was talking about thinking, not optimizing then.

Ok, I was just trying to point out that thinking about optimizing too much during design is premature. I thought thats what you were trying to point out, too.

You are trying to prove a point that isn’t there. Programming is about programmers.

But software projects are about delivering value to your customers, not about programming. Thats why I stopped taking Engineering classes and started taking MBA classes.

It also has to do with how manageable and maintainable a project is over the long run. I have seen too many projects with “amateur” coding became so unmaintainable so fast that they had to be abandoned and had to go for total rewrite.

And sadly, I’ve seen some of those “amateur” projects chug along, working for years while the “professional” rewrite fails to catch up. Re-writing is risky business. Whenever a programmer says “this code is too crappy to change, lets re-write the whole thing,” I change that in my mind to “I’m not talented enough or patient enough to fix this crappy code.” Especially if they are under 30 years old. Hows that for a paradigm shift?

I worked on three generations of manufacturing ERP system over the course of 9 1/2 years. That system was deployed into about 45 somewhat independent installations, replacing several different existing custom solutions. Over the course of this process I was able to see the long term consequences of many technical decisions. I was also able to see several different independent implementations of the same thing. This is where I gained a healthy respect for the ability of “amateur” programmers to write code that successfully meets user requirements. I also learned that technically sound projects can fail spectacularly.

Would it surprise you to learn that one of my last tasks on this ERP system was to begin the migration of the next version to Java? I might know more Java than you think. Then again, I may know less than I think.

I am familiar with XP practices. I deploy my unit tested PHP application weekly, thank you. Interestingly, PHP has a small active community embracing XP principles.

I’m going to address the good example issue in a new post.