Re: short assumed (was: alternates (was: What should go into 1.5.7?))
On Thursday, February 3, 2005 at 11:41:27 PM +0900, Tamotsu Takahashi wrote:
> On Thu, Feb 03, 2005 at 01:50:04AM +0100, Alain Bench wrote:
>> U+FFFD character (???)
Not in your charset, so replacement char itself replaced by three
question marks in my quote. EUC-JP terminal?
> Why does it prevent mutt_convert_string() from masking unconvertible
> chars? Or it doesn't?
I suppose not. With short.3 I see unconvs all masked, just with
different chars (U+003F in body versus U+FFFD in headers). With your
additional mini-patch on outrepl, all is masked by U+003F. Fine.
Otherwise, if you really see unconvertable not masked, please give
us an example, and used settings.
>> it may seem logical to use [U+FFFD] to replace unconvertables, where
>> available, in all sorts of Unicode, in GB18030... Discussable.
It would have to be used for body too. And probably have a user
settable masking char could be interesting. Better a multi-char list,
where the first available in $charset would be used. With something as
"U+FFFD:U+003F" as default.
While at it, when displayed the mask should be colored. And before
deciding to mask an unconv, Mutt could optionally try to transliterate
it if possible (wish #1315). This transliterated char(s) would be
colored on screen. And there would be one mask per unconv char, not per
unconv byte. And... OK, I shutup -h now. Sorry.
> Even if the problem was my fault, $assumed_charset and $file_charset
> are much better than nothing. :) It correctly handles invalid headers
> in almost all cases. (And the current mutt doesn't, at all.) I wish
> they will be added to the CVS code soon.
I agree: Fine today, much needed, only way to solve some cases,
helps much in some other cases, improves user control, has no known
intrinsic drawbacks, no change to default behaviour, indispensable to
UTF-8 users... Numerous problems solved on mutt-users, comp.mail.mutt,
and Mutt, Debian, or Gentoo BTSes since monthes prove it.
It's not just about tolerance to evil mailers, but about working
correctly and controlably in adverse conditions. Real world conditions.
And specific conditions too, see <20031207170127.GA3582@xxxxxxxxx>
on mutt-users for an example about Macintosh converted mailbox.
Bye! Alain.
--
« Be liberal in what you accept, and conservative in what you send. »
Jon Postel / Robustness Principle / RFC 1122