So whatever happened to the whole PGDCE thing?

[JC_]

Why do you say using something else is nonsense?

If you're making a real 3D game so you need complex and standardized tools with some continuity and with good visual/graphics possibilities (in case that you are not John Carmack).
It's not only about people but also about time and money. There are some good non-commercial 3D engines? Or exists more commercial engines which can compete Unity, UE, CryEngine?
I like games like Legend of Grimrock with own 3D engine which have benefits from optimalization etc., but this is another case. They have team of people making engine directly for the game
not not kind of universal thing. Secondly, I have not seen such game with own 3D engine done in Pascal or in Java etc. => if somebody want 100% performance logically he uses the language that will allow it.

I strongly disagree with you on this. Even when rendering 2D graphics you could greatly benefit from features available in newer API's.

Can you specify which benefits and features? OpenGL have shader support since 2004, which was maybe most important thing for 2D graphics rendering?

Why do you think all Pascal engines/frameworks are dead? Because there weren't enough interest in using them. And why weren't enough interest from people in using them? One of the reasons is people always saying of how much they suck instead of trying to help improving them. Pretty much like you now.

If you need more real reason: 90% of them were not usable just some start-up projects. Working systematically is challenging. In second case, capable author(s) lost interest ( own life, no time, growing complexity, limitations of Pascal language..).
You can use same concept again and it will probably fail again. I do not know how you, but I preferred the solution which survived such natural selection and are universal like SDL or SFML..
This increases the chances to finish game in 1 person lifetime.

Why should the posted code be optimized? So that you can only copy and paste it into your project? This way you won't learn much especially because heavily optimized code is often much harder to understand.
I would recommend that instead of worrying about the posted code being optimized we provide as many possible approaches in solving specific problems. This way we allow people to chose the approach which suites them the most. And that largely depends on their coding style and the rest of the project code.

I do not understand much here, eg. ZenGL code is well optimized and very easy understandable and modifiable. On other hand there is ZenGL Particle editor from different author who did great job but it's pure chaos.
Chaos is not optimized in that sense