Ar an cúigiú lá déag de mí Bealtaine, scríobh Stephen J. Turnbull:
Aidan Kehoe writes:
>
> Ar an cúigiú lá déag de mí Bealtaine, scríobh Stephen J. Turnbull:
>
> > In XML, it's strictly enforced by validating parsers;
>
> Right, but there's no guarantee a validating parser has been run on a given
> XML file that XEmacs sees.
Sure. My point is that an XML charset declaration that is wrong is an
error, and by definition other software can treat that case as
deprecated. Users who don't know the rules, well, too bad. But when
there is no rule outside of XEmacs, it's not reasonable to expect the
users to enforce it on their environment.
I want to enforce something I already know on XEmacs, in a way that will
work on other machines that don’t have my value of file-coding-system-alist.
In a way that won’t interfere with command line parsing, as happens on
Cygwin currently:
/usr/bin/env: perl # -*- coding: windows-1251 -*-: No such file or directory
Having it work on GNU Emacs too would be gravy.
> > We don't *need* to do anything; the user can always
go back and
> > re-read the file with C-u C-x C-f.
>
> Oh, don't be obtuse. On that reasoning, we can abandon development
> right now since anyone who would like a feature we don't have can
> implement it themselves.
I'm not being obtuse. Lack of cookie support is not a security or
data loss issue (by itself), nor is it a bug at all; as you point out,
it's a missing feature. We don't *need* to do anything.
Only in the sense I gave above.
So let's take the time to not introduce new ways to fail.
Everything we do introduces new ways to fail. The interesting thing is the
tradeoff between those and the positive aspects of what we do.
If the work to prevent cookie lossage helps deal with our existing
sterile equine behavior, so much the better.
> > No, that is an error mode of sniffing from cookies that we *create*
> > for our users by parsing cookies that are not enforced by the
> > protocol we're editing,
>
> The same error mode arises when the automatic coding detection
> (independent of coding cookies) gets something wrong. It's distinct
> from sniffing cookies.
No, from the point of view of our workflow, it's not the same.
Autodetection failures are in principle our responsibility, and in
theory can be reduced to near zero by Sufficiently Smart Programming.
Spoken like an East Asian. No they can’t. Trivial examples; deciding between
iso-8859-1 vs. iso-8859-15. Between KOI8-U used for Russian vs. KOI8-R.
iso-8859-2 used for German, vs iso-8859-1 used for German (bit-for-bit
identical, until someone types the word Škoda, which Germans have occasion
to do now and then).
OTOH, the rate of broken cookies was quite high in the early days of
Nihongo Emacs; I didn't stop seeing them until cookies themselves
disappeared (in modern MULE, distinguishing among the 4 common
Japanese encodings is essentially perfect). The problem was that
people would do things like modify dos2unix to also recode Shit JIS as
EUC-JP, but not to recognize Emacs coding cookies (a reasonable
behavior for vi users, you will admit, I hope). So you'd get whole
trees full of EUC-JP marked as Shit JIS.
Japan; the land of transcoding HTTP proxies. The land where backslash is
written as Y with two lines through it. The land of government standards for
character sets that left encodings underspecified.
Okay, so we don’t need coding cookies in Japan today. That doesn’t help
Europe a whole lot.
> If you can find it without too much searching, I'd
appreciate a
> link to what Ilya posted. Otherwise I suppose I can ask him.
Will do.
Thanks.
I gather you're not in a "by 10pm tonight" kind of
hurry?
Nope.
(Thing is, I suspect this is in archives that aren't online, may
as
well take the time to put them online.)
--
On the quay of the little Black Sea port, where the rescued pair came once
more into contact with civilization, Dobrinton was bitten by a dog which was
assumed to be mad, though it may only have been indiscriminating. (Saki)
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta