[Catalyst] RFC: The paradox of choice in web development

[Andrew Rodland]

On Thursday 19 February 2009 09:12:36 am Matt Pitts wrote:
> In today's world of software that is cross-platform and OS agnostic at
> its core, Perl 5 is showing its age. Still love it though.
>

This isn't as much a Perl problem as it seems to be -- it's the rule all 
around that writing code that works on _everything but Windows_ is ten times 
easier than writing code that works on everything including Windows. Perl is 
just in a unique place to show this. In C, which is hardly more than a 
portable(-ish) layer on top of assembler, and which has a small standard 
library, code isn't portable at all without significant work (and even still, 
Windows is usually the hardest target to hit.) In Java, portability is 
considered paramount, so OS facilities are exposed through thick 
compatibility layers or else not at all. Perl sits in the middle ground. 
Sufficiently "pure perl" code will run on a million and one platforms, but at 
the same time Perl was never afraid to expose OS facilities (like stat or 
SysV IPC) more or less directly, to allow more powerful code. This has made 
it easy to write code that, even though it doesn't use XS as all, is 
platform-specific enough to crash and burn on windows. But if it's a 
shortcoming in Perl, how do we fix it? By taking all the goodies away from 
the Unix folks so everyone has to write to the least common denominator?