Re: Objective-C standard

[Christopher Armstrong]

**Jeremy Tregunna wrote:

> You guys seem to be missing what a standard is used for. It's not  
> used to keep a language static, it's used to help other people write  
> other compilers and have them work with other compilers. In this  
> regard, you really have to reverse engineer a lot of ruby or python  
> to write a compiler for those languages, whereas with Common Lisp or  
> Scheme, you don't, you have a specification you can follow which  
> explains things in great detail.

I understand the reasons for standardising languages; it certainly helps 
with C++ to get competing compilers to operate the same with a single 
piece of code (although significant issues still exist, like with STL). 
Standards like Intel's C++ x86-64 ABI are also very helpful where 
several different ABI's exist, and where many continue to change from 
version to version. Objective-C is in the somewhat unusual situation 
where it is quite an old language, but IMHO due to a lack of widespread 
popularity and only a handful of compiler implementations, has never 
been standardised. In this respect, having only to target a couple of 
compilers makes it easier to port code, but more to the point, the 
compiler producer can take advantage of this fact and add features to 
the language without differentiating themselves too significantly from 
competing vendors when there isn't many.

Many of the mentioned scripting languages (ruby, perl, python, etc) 
exist in a similar state, and despite being popular, many only have one 
implementation. This means adding features is less harmful to the 
general programming community surrounding this language, as existing 
code bases only have to target the one implementation. OTOH, if MS were 
to add new features to it's C++ compiler (and they have before), and if 
these features were to become popular, they start to create havoc for 
code bases that are in C++ and need to be ported to other C++ compilers 
(like gcc). As programming languages begin to gain more implementations, 
this problem would start to appear.

Regards
Christopher Armstrong