Re: [STUMP] Patch: Non ASCII Chars in Messages

[David Hansen]

On Mon, 06 Nov 2006 09:57:57 +0000 Jim Ottaway wrote:

> David Hansen <address@hidden> writes:
>
>> On Sun, 05 Nov 2006 17:19:56 -0800 Shawn Betts wrote:
>>
>>> Thanks David, your hacking is appreciated :). I've started noticing
>>> encoding errors that I think your patch will fix. I'll test out your
>>> patch next time I get the error.
>>
>> Just `echo-string' with a non ASCII character should trigger
>> the error.  But it would be really interesting if someone with
>> a latin-1 lisp (sbcl w/o unicode support or cmucl should do
>> it) can test it.
>
> I am running stumpwm with cmucl 19c. I have tried to test the patches,
> but I don't know if I am using the correct test.
>
> I tried this:
>
> in stumpwmrc:
>
> (defun test ()
>   (echo-string (current-screen) "∫dϕ = 2π"))
>
> (setq *screen-mode-line-format* "∫dϕ = 2π %w")
>
>
> And then at the repl I did (test) without the patches.  This didn't
> generate an error, but the message from #'test was blank, and so was the
> modeline.  
>
> With the patches applied, still no error, but I did get the latin-1
> rendition of "∫dϕ = 2π" in both the message and the mode-line.
>
> I suspect that this isn't the correct test, though, since I was
> expecting some error before applying the patch. 

I don't quite understand why nothing shows up w/o the patch
but it's expected that some rubbish shows up for "non latin-1"
(i assume the cmucl reader just reads a UTF-8 string as
latin-1).  And i don't think we should try to fix it (just
adding a bunch of code to replace funky latin chars with a `?'
for sequences that look like UTF-8 isn't worth the work,
IMHO).

Please try the patch with just some latin-1 chars > 2⁷: some
Ümläüts or the name of you favorite 80s metal band (take care
you feed them latin-1 decoded to the lisp reader and make sure
you use a latin-1 font).

Another possibility is to test for the features :sbcl and
:utf-8 or :clisp and only use the new translate function in
these cases.

David