Per-process vs per-thread current-directory once more Lassi Kortela (02 Aug 2020 19:36 UTC)
Re: Per-process vs per-thread current-directory once more Marc Nieper-Wißkirchen (03 Aug 2020 07:30 UTC)
Re: Per-process vs per-thread current-directory once more Lassi Kortela (03 Aug 2020 08:04 UTC)
Re: Per-process vs per-thread current-directory once more Marc Nieper-Wißkirchen (03 Aug 2020 12:12 UTC)
Re: Per-process vs per-thread current-directory once more Lassi Kortela (03 Aug 2020 12:20 UTC)
Re: Per-process vs per-thread current-directory once more Lassi Kortela (03 Aug 2020 08:08 UTC)
Re: Per-process vs per-thread current-directory once more Lassi Kortela (03 Aug 2020 08:17 UTC)
Re: Per-process vs per-thread current-directory once more Lassi Kortela (03 Aug 2020 08:25 UTC)

Re: Per-process vs per-thread current-directory once more Lassi Kortela 03 Aug 2020 08:25 UTC

> (parameterize ((current-directory #f))
>    (readlink "foo/bar"))
>
> to temporarily disable the per-thread current-directory for SRFI 170
> procedures and use the OS current-directory.

Wait, #f works in neither Gambit nor Kawa since they type-check the
parameter. It needs to be a string. (Kawa has path objects so it can
also be a path, and a string is turned into a path. The Kawa parameter
is called `current-path`.)

We could have:

(parameterize ((current-directory (use-os-current-directory)))
   (read-symlink "foo/bar"))

where (use-os-current-directory) returns an implementation-defined
object that skips the per-thread current-directory.

Apparently it also works as expected to make an alias for a parameter:

(define current-directory current-path)