Re: Annotations for command line arguments and environment variables

["Jasper St. Pierre" <jstpierre mecheye net>]

Short answer: "filename" seems correct to me. "filename" really is a
poorly named term that means "the current locale that filenames and
terminal text is encoded in". So, if you set your locale to
LANG=ru_RU.KOI8-R or LANG=ja_JP.SJIS, then text in your command line
should be encoded as such. If GTK+ really wants to handle arbitrary,
non-encoding-supported bytes in its command line arguments, well, it
can, I just think that's a really poor idea, especially in
higher-level languages.

Longer answer: The issue is that POSIX and Python 3 have different
views of the world.

POSIX says "bytes, bytes, everything is bytes. Do Unicode internally
if you really need it".

Python 3 says "Unicode please, bytes are for low-level system
hackers". Python 3 does not have a proper bytes string type. So,
Python's sys.argv is Unicode. It receives bytes through the C API, and
then decodes that internally using the same LANG environment
variables. PyGObject receives Unicode strings, and needs to figure out
which bytes to pass in a char*. GI gives you two options: use UTF-8
always. "filename" is the correct choice to convert Unicode strings
into an argv.

On Tue, Jun 7, 2016 at 10:28 AM, Christoph Reiter
<reiter christoph gmail com> wrote:
> Hey everyone,
>
> I'm currently trying to improve filename and argv/environ handling in
> PyGObject and started annotating filenames as "filename" and also argv/environ
> related arguments as "filename" as they are treated the same way in Python.
>
> I've (among others) filed
>
> https://bugzilla.gnome.org/show_bug.cgi?id=767266
>
> and
>
> https://bugzilla.gnome.org/show_bug.cgi?id=767268
>
> where it was pointed out that argv/environ should maybe not be annotated as
> filename and on IRC uint8 arrays were suggested instead. The later would
> result in API breakage in PyGObject, but could be mostly worked around using
> overrides.
>
> So, I'm wondering how other bindings handle that issue and how an annotation
> change for, say gtk_init() to either "filename" or "uint8 array" would affect
> other bindings. Or if someone has a better idea.
> _______________________________________________
> gir-devel-list mailing list
> gir-devel-list gnome org
> https://mail.gnome.org/mailman/listinfo/gir-devel-list



-- 
  Jasper