[buug] recode vs. iconv

[Ian Zimmerman]

Turns out there a whole thicket of issues here.

First, I lied when I said using iconv was the solution.  It converted
_most_ of the latin1 letters but not some of the more exotic ones, like
"Đ".  It replaced the latter with question marks :-(  As the program
(and the underlying part of libc) was written by Uli Drepper I wonder if
it just converts the characters used in German and gives up on the
others.

I also lied when I said I read the whole recode manual.  When I went
over it I had already read the debian-written manpage, so I skipped the
Invoking section where the options are documented.  Fail!  The manpage
is out of date.  It omits the -x option, and it doesn't mention the
"unification" of recode and iconv which is rather important.  This:

recode latin1..ascii < foo.txt > /dev/null

fails on the first non-ascii character, while this:

recode -x: latin1..ascii < foo.txt > /dev/null

"succeeds" - but then it is a rather slower version of cat.

I think this unification happened after your experience with recode.

The only package in debian that did something really close to what I
wanted was cstocs.  I like its Czech-specific features too :) so I am
glad I have learned about it.

But we sure have cooler icons and transparent windows now :(

-- 
Ian Zimmerman
gpg public key: 1024D/C6FF61AD
fingerprint: 66DC D68F 5C1B 4D71 2EE5  BD03 8A00 786C C6FF 61AD
http://www.gravatar.com/avatar/c66875cda51109f76c6312f4d4743d1e.png
Rule 420: All persons more than eight miles high to leave the court.