Re: Compiler for Red Hat Linux 8

[Russ Allbery <rra at stanford dot edu>]

Bernard Dautrevaux <Dautrevaux@microprocess.com> writes:

> Looking at the headers you'll see that you send only us-ascii (a quite
> old subset of iso-8859) while a lot of people use the *standard* iso
> charset, where 0x91 is "opening single quote", 0x92 closing, 0x93
> "opening double quote" and 0x94 closing.

No, they're not.  ISO 8859-1 doesn't assign any code points between 0x7F
and 0xA0.

0x91 through 0x94 are assigned that meaning in Windows Code Page 1252,
which is not a standard of any type whatsoever.  Those characters are used
for Windows "smart quotes", and will be unreadable to anyone not using
Windows unless their software specifically caters to mislabelled character
sets or the message is correctly labelled and there is a CP-1252 font
available.  Messages using those characters should be properly tagged with
charset=windows-1252 or they're in violation of the MIME standard.  Most
software on Microsoft platforms does not correctly label this character
set and produces broken messages like the one that started this sub-thread
(which was labelled as ISO 8859-1, despite the fact that it didn't use
that character set).

I'm sure the original author was completely unaware that this was
happening; most Microsoft software makes it extremely easy to make this
mistake and apparently doesn't care about producing readable messages on
non-Microsoft platforms.  Usually turning off Smart Quotes will fix the
problem.

<http://ppewww.ph.gla.ac.uk/%7Eflavell/iso8859/iso8859-pointers.html#unass>
has additional details on why those code points are intentionally
unassigned in ISO 8859-1 (on 7-bit systems, they could be interpreted as
control codes when the eighth bit was stripped, producing serious
problems -- this was a significant issue at the time that ISO 8859-1 was
standardized).

-- 
Russ Allbery (rra@stanford.edu)             <http://www.eyrie.org/~eagle/>