Re: Unicode-based FreeBSD

Erik,

2008/8/23, Erik Trulsson <ertr1013_at_student.uu.se>:
>
> There are many applications that do not yet support UTF-8.
> It would be bad if applications that just output 8-bit characters "as-is"
> were broken.
> If an application were to output characters from (e.g.) ISO-8859-1 and
> syscons were to interpret them as UTF-8 it would not be pretty.
>
> I suspect it would actually break many current applications.
>
I agree that the proposed solution will have no effect on pure ASCII
applications and would break apps that generate high bit characters of 8-bit
encodings. My ideas on that are:

1) I mostly use FreeBSD in character mode with pure ASCII applications. For
web browsing, writing e-mails and similar tasks I use X-based applications
that have their own charset handling.

2) Adding the ability to map from an arbitrary 8-bit encoding (i.e. just
keep the current features) is not hard.

3) Fixing the subset of applications that work in character mode and
actually generate 8-bit characters is doable.

Please note, that UTF-8 was specially designed for full interoperability
with ASCII and partial with 8-bit encodings. For example, if we have an
application that just performs a search for string of bytes in its input, it
will work equally well if given iso-latin1 text and if given UTF-8 text.

The real-life example is vi. Once I realized that kdm reads full user name
as UTF-8 and that my FreeBSD is using koi8-r, I just took konsole, switched
it to UTF-8, started vi and edited /etc/passwd as if it was UTF-8 (it
actually was pure ASCII). And after that I am able to see correct russian
names of users on my home PC in kdm window.

So if someone thinks that many apps would be broken, let's name a few and I
will test them using konsole and UTF-8.

And again, how to check out the source, what is correct branch/tag? Should I
check out from CVS or svn? To my mind, if I modify source code locally this
certainly would not break applications on other FreeBSDs in the world. :-)

Alexander Churanov