Re: The case for glob in SRFI 170
Lassi Kortela 11 Dec 2019 21:00 UTC
We have exactly the same arguments :)
> support of "**"
This is probably the most important missing feature in traditional
globs. Without "**" you have to know precisely how many levels deep you
want to nest and write "*/*/*" manually.
It's like those very old Unix regexps that are missing the "." wildcard
or some such. You have only "*".
> - Glob-like matcher is sometimes useful outside of traditional
> filesystem.
Good idea. This might fit a typeclass for tree traversal. We probably
can't make zippers work?
<https://en.wikipedia.org/wiki/Zipper_(data_structure)>
> (To be precise,
> posix does provide alternative subsystem via GLOB_ALTDIRFUNC, but to use
> that from Scheme there would be lots of C-Scheme hopping needed.)
This is a good example of the kind of problem that libc causes with
high-level languages. It can be better to implement the same thing from
scratch in the HLL. The C interface would have so many workarounds that
it adds up to the same amount of code.