> > Package: xlibs
> > Version: 4.3.0-2
> > Severity: normal
> >
> > This is very annoying! I can no longer get a quotedbl (") by pressing
> > <dead_diaeresis> <space>, as it was always. Why this change? A
> > diaeresis can still be produced with <dead_diaeresis>
> > <dead_diaeresis>, should anybody need it (!?).
> >
> > I suggest you to apply the following patch. I hope there's no plans to
> > change <dead_acute> <space> to produce an useles <acute> instead of an
> > apostrophe... Why do people change things they don't use??
>
> I am sorry, but I am reluctant to break with upstream practice on this
> issue without understanding the underlying issue a little better.
Well, what can I tell you...
There are people who have a keyboard with US layout, but need to type
non-ascii latin characters. For example, Germans living in the US will
likely have an US keyboard, but they still need to write umlauts when
they write in German. Or people living in countries where keyboards
are imported... sometimes keyboard availability depends on
internacional prices, and there are times when you can only get an
US keyboard, other times only an es-ES keyboard, and others only the
ill-designed es-LA one.
Anyway, these people use an US keyboard and need to input diacritics.
The standard, intuitive and historical way is to use the keys labeled
with symbols resembling the diacritics to input them. Thus, pressing
the apostrophe key produces a <dead_acute>, the quote a
<dead_diaeresis>, the circumflex a <dead_circumflex>, etc. The next
pressed key will result in an appropiately diacriticalised(?) letter.
So far, so good. But how to input the apostrophe, quotedbl, and
asciicircum? The historical way was to press the key followed by a
space. This is was they broke with the new en_US.UTF-8/Compose.
Why <dead_diaeresis> <space> should produce <quotedbl>:
1. Ability to produce quotedbl. When these people decided to "improve"
the rules file, they took out the iso9995-3 symbols from the layout
"pc*", variant "us_intl". So, people using UTF-8 locale, and using
that layout/variant combination can no longer type a quotedbl
(us_intl maps the quote key to dead_diaeresis).
2. Historical/backwards compatibility. Compatiblity with *me*, the
dead_diaeresis user. I'm used to type a quote that way. I'm so used
to it that I always type '" ' when I use a pure US keyboard layout,
and need to backspace. It's frustrating to have one's keyboard
change its behaviour according to the academic theories of the last
XKB commiter.
3. Compatibility with other "pseudo-diacritical" keys. I'm afraid of
saying that, should they inutilise their composing sequences too...
<dead_acute> <space>: apostrophe, <dead_tilde> <sp>: asciitilde,
<dead_grave> <sp>: grave.
4. It doesn't make sense. Maybe these people thought: "We were doing
it wrong all the time! <dead_diaeresis> + space should produce a
diacriticalised(?) space! That's the logical thing to do!" Well,
there is no such thing as SPACE WITH DIAERESIS, and that key
combination only produces a DIAERESIS. Ok, their reasoning is not
too far-fetched, but there's already a way to produce a lone
diacritic: pressing the key two times; and see next item.
5. WHY WOULD ANYBODY WANT TO PRODUCE A DIAERESIS? I can only think of
one context: when you are talking about diaeresises. Like: "The
diaeresis (¨) ...", or "<dead_diaeresis> <space> produces '¨'!".
The ratio of usage of quotedbl vs. diaeresis, I think, must be
infinity to one. Even if you have a keyboard with both a quotedbl
and a diaeresis key, if you press <dead_diaeresis> <space>, the
most likely thing is that you want a quotedbl. I think Windows
applies the same line of reasoning that the XKB people, because
many times I have seen mails in which the sender quoted ¨like this¨
or ´like this´. They are obviously searching for the quote or
apostrophe keys, and settle for the nearest similar-looking one. Of
course they don't want the diacriticals but the quotes, because ...
(see beginning of paragraph).
As an addition, I think the compose sequences should be consistent
across locales, but maybe I'm mistaken. And, in any case, I think the
"consistency trend" should be in the opposite direction (there are
already 9 locales with the wrong composing sequence (en_US.UTF-8 is
modified by me)).
$ grep "^<dead_diaeresis> <space>" */Compose | expand -t1
el_GR.UTF-8/Compose:<dead_diaeresis> <space> : "¨" U00a8
el_GR.UTF-8/Compose:<dead_diaeresis> <space> : "¨" diaeresis
en_US.UTF-8/Compose:<dead_diaeresis> <space> : "\"" quotedbl
iso8859-15/Compose:<dead_diaeresis> <space> : "\"" quotedbl
iso8859-1/Compose:<dead_diaeresis> <space> : "\"" quotedbl
iso8859-2/Compose:<dead_diaeresis> <space> : "\250" diaeresis
iso8859-3/Compose:<dead_diaeresis> <space> : "\250" diaeresis
iso8859-7/Compose:<dead_diaeresis> <space> : "\250" diaeresis
iso8859-7/Compose:<dead_diaeresis> <space> : "\250" diaeresis
iso8859-9/Compose:<dead_diaeresis> <space> : "\250" diaeresis
iso8859-9e/Compose:<dead_diaeresis> <space> : "\250" diaeresis
pt_BR.UTF-8/Compose:<dead_diaeresis> <space> : "\"" quotedbl
But I won't protest for a locale I don't use. en_US.UTF-8 is the
important for me.
I hope these notes were helpful to you.
Bye.
