[PD] Windows in Chinese, problem with paths

[IOhannes m zmoelnig]

On 10/22/20 3:36 PM, Maximiliano Estudies wrote:
> Dear list,
> 
> I am working on a project with people in China and sent them a project with
> a main patch and a couple of subfolders for abstractions and audio files.
> I'm using [declare -path abstractions] so PD finds the files relative to
> the main patch.
> I tested the patch (I copied the same .zip file that I'm sending) in an
> older windows machine I have and it works fine, but the chinese colleagues
> are having some trouble. They sent me a screenshot of the console and PD
> can't find any of the subpatches or audio files.

could you also share this screenshot (at least the important parts, 
without disclosing confidential information)?


> Do you have any experience with this? They are using Pd 0.51.2 on a Windows
> 10 machine with the operating system in chinese. I checked and they didn't
> change the folder structure.

like most people here (judging from the lack of responses to your email) 
i have little experience with chinese systems.

the first thing that comes to my mind when reading your problem 
description is a character representation issue.
Pd uses UTF-8 (a Unicode variant) to represent characters.
this is pretty standard in the entire world, except for Windows where 
UTF-16 (if they use unicode at all) is used.

i think Pd is pretty well tested for doing a proper conversion between 
it's UTF-8 representation and the wide-char (UTF-16) representation 
needed on Windows - but these tests are mostly done on "western" 
systems, where the characters are practically always formed by 
unicode-points <65536 (which is the highest number that can be 
represented by an (unsigned) 16bit value; UTF-16 uses 16bit (hence the 
name) as the base number size).

afaik (but - see above - i know little) the standard code points for 
chinese letters are in the range of 0x4E00..0x9FFF, so they would well 
fit into the 16bit constraint.
but then: you never know what encoding they really use.

so:
- could you share the screenshot?
- would it be possible to find out which character encoding is used by 
your colleagues? and which codepage? and...? i'm not on windows but a 
quick websearch revealed [1], which has at least a bit of information on 
how to get all these things)
- would it also be possible to get the full path to one of the failing 
abstractions as *text*? (not as a screenshot).

gfmdasrf
IOhannes

PS: another thing: if you are using externals that open files by 
themselves, they really *should* use Pd's file-opening functions if they 
want to open non-ASCII filenames.
so if the problematic files are not opened via externals, it could well 
be a bug in the externals (though from your description i figure that Pd 
itself is having problems)



[1] https://serverfault.com/questions/80635/