Master thesis

[Patrick Schluter via Digitalmars-d]

On Saturday, 1 July 2017 at 15:58:04 UTC, ketmar wrote:
> Dgame wrote:
>
>> On Saturday, 1 July 2017 at 14:40:36 UTC, ketmar wrote:
>>> Dgame wrote:
>>>
>>>> Which impact would have D on the software-architecture, if 
>>>> it would be choosen for a 2D game instead of C/C++?
>>> i can actually finish 'em. most of the time when i'm working 
>>> with D, i feel that compiler tries to help me. EVERY TIME i'm 
>>> working with C/C++, i know for sure that compiler silently 
>>> setting traps for me ('cause it is absolutely impossible to 
>>> write safe and secure C/C++ code
>> Do you have examples (code and/or personal experience) about 
>> that?
> basically, all of my projects. ;-) like zx spectrum emulator, 
> which took monthes in C, and only weeks in D (and it almost on 
> par feature-wise with C version, and even more advanced in some 
> areas like FDC emulation).
>
>>> -- tnx to standard committee and compiler implementers, 
>>> anything can turn into UB unexpectedly, and writing UB-free 
>>> code is tedious and it will have UB anyway).
>> Is that described somewhere? Or do you have examples or both?
> ah, many people are complaining. just recent ones (there are 
> MUCH MORE out there, just follow the links in the articles, for 
> example):
>
> http://www.yodaiken.com/2017/06/26/the-c-standard-versus-c-and-the-mother-of-all-hacks/
> http://www.yodaiken.com/2017/01/05/undefined-behavior-and-the-purpose-of-c/
>
> everything compiler authors has to say is (see gcc bugzilla 
> link in one of the articles): "it is permitted by the standard. 
> don't write code with UB". that is, modern optimizing C 
> compilers are *actively* trying to *subvert* your code. using 
> modern C compiler ends up in you being punished for all kind of 
> random things: compiler is using *any* your error against you. 
> it probably helps to win some stupid compiler benchmarks, but 
> it is crystal clear that modern C is not made to write 
> real-world programs in it: programmer have to manually track 
> the things, and make machine happy. this is exactly the 
> *opposite* of the machine's purpose: it is machine's task to 
> make human happy, and free human from tedious and error-prone 
> work, not vice versa! ;-)
>
> besides those issues, the GREAT advantage of D is well-defined 
> low-level types. `uint` is always 32 bits, `short` is always 16 
> bits and 2-complement (and it has overflow defined!), and so 
> on. no need to `stdint.h` anymore (and why do we need it in C 
> at all? winning several CPU cycles on the occasion vs hours of 
> fighting to write portable code... OBEY THE MACHINE, YOU 
> WORTHLESS HUMAN! ;-)
>
> i dropped C years ago, switched to D, and i will NEVER return. 
> D was made by humans, for humans, not by committee and compiler 
> writers to win useless benchmarks.

Amen.