[Ecls-list] Latest changes

Juan Jose Garcia-Ripoll jjgarcia at users.sourceforge.net
Tue Nov 14 14:19:45 UTC 2006

2006/11/14, Yaroslav Kavenchuk <kavenchuk at gmail.com>:
> Small offer: when ECL build with unicode do not use (from CHANGELOG)
> on the ISO C99 library functions for wide characters" - this is not work
> correctly for non-ascii characters on win32 (on linux?). Other library
> is necessary (instead of I know it) or own functions, based on
> UnicodeData - http://unicode.org/Public/UNIDATA/UnicodeData.txt (as in
> sbcl/clisp).

Let's go step by step, ok? ;-) Right now what there is works with GNU
libc but the functions are encapsulated (ecl_alphanumericp(),
ecl_char_upcase(), etc, etc). It should be easy to replace these
functions with a different library for handling unicode characters,
given that you have one. But that is currently not my priority,
because there are more important things to fix first -- streams, as I

> > What is left is probably the hardest thing. Namely, I/O with Unicode
> > characters. This must necessarily be accompanied by a redesign of
> > ECL's streams which allows for extensible streams, external formats
> > and buffers handled by lisp.
> Maybe use libiconv? (it is on win32 too)

libiconv is only for translating strings. It is not thought for
handling character i/o, or at least it is not trivial to me how to use
it for reading data from an octet based stream, since it does not have
functions to determine whether a set of octets forms a valid character
or not.


More information about the ecl-devel mailing list