Le jeudi 27 octobre 2005 à 21:10 +0200, cono a écrit :
> Nicolas Mailhot wrote:
> 
> > Le jeudi 27 octobre 2005 à 14:12 +0000, Andrew Brown a écrit :
> > 
> >>
> >>This could surely be cludged around with an addin. I know and you know that 
> >>the underlying mechanism would still be styled, but no one else would :-)
> > 
> > 
> > If only it was that simple :(
> > 
> > If you don't also neuter the UI that enables setting language in styles,
> > you'll have collisions between styling and your new langage control,
> > with users asking why the styles are suddenly changing the language they
> > specified when typing their text.
> > 
> > I do agree that a better cludge than the current cludge is possible, but
> > till it's uncludged it'll have cludgy side-effects.
> > 
> 
> I do recognize the idea of Andrew. And as far as I understand it, it 
> won't conflict with the style feature.
> 
> Sometimes it is like you do want to keep people away of the effort to 
> learn about styles, Nicolas.
> For me there's no misundertanding about styles and what the language 
> does with it. And so many more things about styles are valuable to know. 
> For ease of work, to prevent format-mess, ...

I'm not opposed to styles. Far from it. But I was exposed to the
concepts behind styles way before I was exposed to SO/OO.o, so for me
OO.o styles are just one implementation of this concept. And I'm able to
recognise the shortcomings of this implementation when I meet them. If
you get past the "styles are cool" stage you quickly find ways the
current implementation could be enhanced.

> BTW, and not so unimportant ;-) we've been drifting far far away from 
> the question of the OP: he/she had some troubles with spell checking of 
> different languages in one doc. Because of minunderstanding, or because 
> of a bug?

Actually, we've not strayed so far. The user had trouble with spell
checking of different languages in one doc because the language
implementation in OO.o is utterly alien to him. The enhancements he
proposed are made problematic by this very same implementation. If I had
to guess OO.o language handling happened this way:

1. marketing person tells developer team they have to handle multiple
languages in documents
2. developer team asks if language is some sort of text attribute
(people working with human languages go to different universities than
people working with computers, so they're not sure)
3. marketing person answers yes of course (he got other things to do)
4. no one does any usability study
5. developer team puts language in the style structure with the other
text attributes
6. a user discovers langage in styles enables some sort of langage-based
formatting
7. everyone cheers and congratulates one another on the decision to put
langage in styles

The problem being langage is a text attribute all right but it's an
immutable attribute. It's ok to change the styles of bits of text when
restructuring a document but langage must pretty much be left alone
during these operations (OO.o can't translate text on the fly). 

So any normal human being not formatted by the way OO.o handles language
is confused by the current interface (as any usability study will show).
This user was no exception. And sure when you understand the way OO.o
does styles you can sort of make it work but you have to fight the tool
all the way. It's not behaving the way it should.

I spent ~20 years in the same flat as a professional translator and I
know pretty well what these people expect. Their requirements are way
higher than those of students wanting to put a few latin quotes in their
work assignments. And they're not the same people that do professional
formatting. Asking them to use the same controls is a big mistake. The
whole point of styles is to separate presentation from content and here
you put content elements in the presentation layer (actually there
should also be a document structure layer, but that's another problem)

Regards,

-- 
Nicolas Mailhot

Reply via email to