[Ecls-list] Problems building HEAD with ECL_UNICODE on windows

Gabriel Barta gbarta at gmail.com
Wed May 27 15:46:24 UTC 2009


On Tue, May 26, 2009 at 2:03 AM, Juan Jose Garcia-Ripoll <
juanjose.garciaripoll at googlemail.com> wrote:

>
> What do you mean by broken? That I/O of Unicode streams does not work,
> that text is not displayed properly, ECL fails to boot?


Sorry, my claim was much too strong. Reading unicode from files seems to
work (utf8 at least), it turns out it is just working with the interactive
console which is a problem.  ECL appears to be interpreting the console
input as utf8, but sadly it is not. I think that the characters sent in from
typing in the cmd.exe window are based on the active code page CP_ACP.  So
in that sense stdin should not be interpreted as utf8 but using CP_ACP
instead.  However, if you use shell redirection to send a utf8 file to
stdin, then it is sent in binary form and thus handled correctly.  I am not
sure if there is a way for ECL to distinguish between these two uses of
stdin. :/

<..snip CLOS problem..>

> Uf that is strange. Does it always happen? I have not seen it with
> current sources, but will check again with the flags you used.


Yes it seems to always happen.  It can be reproduced building with the free
VC express 2008.  Because it happens at ecl boot, the only debugging option
seemed to be to debug the c code generated from the lisp sources. And I
couldn't work out what was going on in the generated code...
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.common-lisp.net/pipermail/ecl-devel/attachments/20090528/a5e8283a/attachment.html>


More information about the ecl-devel mailing list