Re: multiple attachments from command line
On Monday, June 13, 2005 at 2:58:41 PM +0200, Ionel Mugurel Ciobica wrote:
> Well, ISO-8859-2 and ISO-8859-16 are almost the same.
Ahum? There are "only" 35 differing chars, plus 16 chars in common
are coded differently. Total 51 differences.
> The reason I use ISO-8859-16 is that the _WRONG_ glyphs S|s|T|t with
> cedilla will show up as _GOOD_ glyphs S|s|T|t with comma below.
I see. You mean Romanian language uses really commas below, but in
practice most people use cedillas as a suboptimal approximation, also
because default ro_RO charmap is (was?) Latin-2 and doesn't contain
commas below. Right?
One problem: Latin-10 in turn doesn't contain cedillas, so Romanian
L2 mails would appear with those characters masked by question marks.
Unacceptable. Thus you feel necessary to do some form of L2 to L10
aliasing (looking at L2 with a L10 font). This solves the problem, but
creates other ones. You don't see or don't care about those other
problems. Right?
And if you see, quote, and type commas, you really send cedillas.
Have you explored iconv //TRANSLITerations? If it is doable, this
would probably be a way better solution than harmfull aliasing. More
natural, cleaner, more controlable, friendly with other term charsets,
and probably without a single drawback.
But it may mean having to modify or expand transliteration table,
and rebuild maybe libiconv, locales, or even the full Glibc. With
libiconv 1.9.2 it would be dead simple: 4 lines to add to
lib/translit.def, make -f Makefile.devel, then normal make. With Glibc
iconv: I don't know, never done it, but it's locale dependant.
> I change easily the view [charset] using the Ctrl+Right_click
And of course you change the locale accordingly?
> change between Latin9, Latin2, Latin10, koi8-r and UTF-8 (which
> actually would be Latin1 here).
What do you mean "would be Latin1"?
> Little mail I get from them is either latin1 or UTF-8 and in both
> cases the header Content-Type is "Content-Type: text/plain;
> charset=us-ascii"...
This case, and "Content-Type: text/plain" without label case, are
both workaroundable with:
| charset-hook ^us-ascii$ iso-8859-1
But you might prefer to alias Ascii to Latin-2 or another. Or do
that from inside a folder-hook.
> I only have problem with yahoo mail which I get in Romanian, which is
> both text/plain and text/html, and the text version contains the http
> sequence of the non-ascii characters without the &. Example: ă (a
> breve) is ă in the text/html and #259; is text/plain... So I have
> to edit the message if I want to read it, or to use the html part.
Yahoo!Mail is known to behave differently depending on the webmail
interface language the user selects in the preferences. Perhaps...
Otherwise a $display_filter might correct the display.
> I made latin1, latin10 and latin9 aliases of latin2. then I can use
> the Latin10 fonts to look at it.
How have you made the aliases?
Such wide range aliasing of uncompatible charsets seems to me a very
bad idea. Whatever brokenness it is intended to workaround, one should
not sacrifice display and quoting of legitimate mails.
> CP1250 indeed has some extra glyphs, but for Romanian language those
> are just quotations
Well yes: 1250 compared to L2 has 27 extra chars, mostly quotes and
other signs including the €uro symbol. But there are also 15 common
letters coded differently.
> [1250 quotes] show as Ctrl-Meta-V or something like that (~V).
Strange: They are supposed to be masked by question marks, or
octalised as say "\222" for the infamous smart apostroph (’).
Bye! Alain.
--
Hotmail users break umlauts for everyone else on a mailing list!
They should stop doing so immediately!
« MSN considered HARMFUL » PCC CB on MU. © June 2002