Re: [Chicken-hackers] [PATCH] Make chicken-install create the destination folder

[Mario Domenech Goulart]

On Mon, 25 Apr 2016 22:30:08 +0200 lemonboy <address@hidden> wrote:

> I was wondering why my fresh chicken5 install was acting up only to
> discover that when initializing the eggs repository using
> chicken-install's -i switch I had supplied a non-existing folder; the
> tool then ends up copying all the files, one over another, in the path
> specified by the user.
> Attached is a patch against the chicken-5 branch that fixes the problem.
>
> From b3b7cec9a18360bf6c1aa9904076f4c62701613b Mon Sep 17 00:00:00 2001
> From: LemonBoy <address@hidden>
> Date: Mon, 25 Apr 2016 22:23:41 +0200
> Subject: [PATCH] Create the destination folder for the -i command
> To: address@hidden
>
> ---
>  chicken-install.scm | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/chicken-install.scm b/chicken-install.scm
> index a9e25ba..9d9fcd9 100644
> --- a/chicken-install.scm
> +++ b/chicken-install.scm
> @@ -230,6 +230,8 @@
>            (copy (if *windows-shell*
>                      "copy"
>                      "cp -r")))
> +      (unless (directory-exists? dir)
> +        (create-directory dir))
>        (print "copying required files to " dir " ...")
>        (for-each
>         (lambda (f)

Nice catch.  However, I think it'd be better to just call
(create-directory dir #t).  It's shorter and handles nested non-existing
directories.

On the other hand, `create-directory' is a bit broken, in my opinion
(but that's another issue -- maybe for CHICKEN 5).  It's subject to race
conditions and will happily succeed if you give it an argument which is
any filesystem object to which file-exits? return a string (e.g., a
regular file).  In its defense, this behavior is documented.

All the best.
Mario
-- 
http://parenteses.org/mario