Inside Starcraft

CUnit < CDoodad < CFlingy < CThingy
Turnkey level ENTERPRISE C++ class hierarchy!

http://www.codeofhonor.com/blog/tough-times-on-the-road-to-starcraft

INSIDE YOUR ANUS

[enterprise]THE FORCED USE OF SEEPLES[/enterprise]

I do that in Java all the time


Vector4f < Vector3f < Vector2f < Vector1f < OneDimensionalPoint

>>1
I doubt they had much choice back in 1996. Home Computers just didnt had enough memory to run Common Lisp. It would have required running it in real mode and doing constant swapping.

>>4

Gross!

groupElement < vector < OneDimensionalPoint
groupElement < vector < Vector1f
groupElement < vector < Vector2f
groupElement < vector < Vector3f
groupElement < vector < Vector4f

projection: Vector4f -> Vector3f
projection: Vector3f -> Vector2f
projection: Vector2f -> Vector1f
projection: Vector1f -> OneDimensionalPoint

That was SEPPLES quality!

StarCraft units, which as mentioned previously contained scads of pointers in the fields for linked lists, was an entirely different beast. It was necessary to fixup all the link pointers (taking special care of unioned pointer fields) so that all 1600 units could be written at once. And then unfixup the link pointers to keep playing. Yuck.
This Patrick Wyatt is an idiot. MMUs allow mmaping (http://en.wikipedia.org/wiki/Mmap) the dynamic (playfield/units state) and static areas (sprites/textures) of the game, so one can easily save/load memory ranges.

I think Blizzard is still afloat, because other programmers are even bigger idiots.

>>8
Now go back 17to years child

It surprised me that they messed up working with doubly-linked lists.  According to the author, every time they had to manipulate the list they did the pointer acrobatics by hand.  This is definitely a WTF moment because function call overhead in 386 family back then was not so severe, and compilers already knew how to inline shorter functions, but even if all of the list mutation occurred inside tight loops and they did not trust their compilers to inline functions where it was really needed, they could just #define some CPP macros (like they do in Linux kernel all the time).  OK, class hierarchy and other architecture issues can be hard even for seasoned programmers, but trusting people who can mess up things as simple as working with well-known data structures is like trusting maintenance of your house to a person who cannot tell a wrench from a hammer.

TELL MY ANUS

>>10
If i read correctly, at first they were using an excellent doubly-linked list library from STORM.dll or w/e, then the newfag asshat developers didn't like it because it made them write more code at certain points so they got rid of it. Then when Patrick Wyatt was the main bug fixer he tried to convince them to go back to it because the newfags couldn't code for shit, but he was denied because the game was launching in 2 months (not really though, i trol u blizzard style)

>>12

I figure Patrick Wyatt was a talkative faggot, he didn't explain why the original developers did as they did, not using STORM doppler linked list shit. I mean there must be some huge ass issues if a wrapper for a linked list is fucked up, how is that even possible? Then there's all the comments from past developers. No one noticed how they're still kissing his ass? As if they're still making sure they're not challenging his view on this subject?

>>13
i didn't read the comments, i'll do that later today

All the SEPPLES devs were angry after part 2 in the comments.

OOP actually gets me hard.

NIGGERS

^* AFRICAN AMERICANS