Perl, et al may have been done right, but it's myopic to think they are some sort of end point of perfection... I'm not sure anyone will argue that seriously.

Perl, et al may have been done right, but it's myopic to think they are some sort of end point of perfection... I'm not sure anyone will argue that seriously. That's an amazing assertion for which I would like to see proof. It's probably not true for Perl 5's concurrency and parallelism features, such as they are. (You get fork() and POSIX-style shared memory with IPC or a queue or RPC or whatever, and you like it.) It's probably true for access to POSIX and other standards. (I can't remember the article here several months ago about how to write fast network code, but it's essentially "Embrace what Unix and Unix-like systems give you." It's not the article about the Varnish architecture.) It's almost definitely true for module repositories, where it's not enough to track dependencies and mirror downloadable code bundles. Testing, change management, dependency resolution tracking, and documentation formats and presentation systems are essential. Not even CTAN gets it as right as CPAN.