mutable languages [was Re: Vectors as functions]

[address@hidden]

> The problem is that many of the interesting things you'd want to
> allow are tied to the data, rather than to the lexical scope the data
> lives in.
>
> You can generally get away with major changes in available behavior
> if you cross over a significant boundary--usually changing languages
> is sufficient warning that the rules will change...

    That's the traditional way to handle the need for different capability
sets, and it has its advantages, but certainly some practical
drawbacks as well.  It must seem daunting to many programmers and / or
managers to add a whole new language to a project to access a few or
even just one feature that language provides beyond the default
project language; it must be tempting to just muddle through with ad
hoc capabilities implemented in the less suitable language.  I would
think in some cases it would be a win to have the ability to merely
spawn a new module in the same language but with enhanced
capabilities.

    And not without drawbacks like keeping track of what is available in
the particular module you're working on.  But in a project where
individuals usually work on certain modules in a certain time frame,
it might not be a problem.