[LINK] Microsoft is dead

Paul Brooks pbrooks-link at layer10.com.au
Tue Apr 10 11:12:00 AEST 2007


Rick Welykochy wrote:
> iTunes and iPhoto rely on symbolic (or perhaps hard) links on Unix
> to spin their magic. The links allow multiple instances of files to
> exist in the file system without duplicating the file space.
>
> Unfortunately such links are not available in Windows* and thus
> to simulate the native Unix behaviour, actual copies are made of
> multimedia files. A gross waste and a very bad deisgn decision, imho.
Or, it could just keep a record of all the actual locations by path, 
without creating thousands of symlinks let alone copying actual files.

> You must admit, though, that different languages support different
> paradigms. It is probably way too much work to attempt a you-beaut
> O-O system in FORTRAN 77 for example. Even trying to write neat and
> clean object *based* code in C can hardly match the elegance (some
> would say nausea) of C++.
>
> If a computer language does not support a certain paradigm, it is
> usually a wasted effort attempting to simulate that paradigm.
No - I usually found it was a function of whether a certain programmer's 
brain supported different paradigms, rather than anything intrinsic in 
the language itself.
Yes, some things are much easier in a language with explicit support for 
those things - but if the developer's brain doesn't understand the 
concept first, it won't be used, or won't be used correctly.
(If your brain doesn't fundamentally understand recursion and stacks, 
writing something in LISP won't help)
For instance, I still don't regard building a malloc/calloc/free module 
into a lump of assembler code as wasted time, it certainly beat the hell 
out of assuming infinite static storage.

> Example: have you ever seriously tried to use an exception handling
> system (throw,catch using macros ... urgh) in C that accurately
> tracks heap-created objects and unwinds the stack properly? Insane!
> So error prone and dangerous that I would never use what I was
> experimenting with in production code. Whereas in C++, perl, python,
> and similar languages, exceptions are used all the time for error
> processing.
sure - but if first, you cannot grok event-driven or non-linear 
execution, then language support doesn't help.


(and recalling that quip about the lurker resurrecting a discussion six 
months on - I think I recall a link discussion about the word 
'paradigm', very similar to the recent discussion about 'space'.
Certainly 'paradigm' is a bigger 'shudder word' to me than 'space' )

(duck)




More information about the Link mailing list