Uwe Stöhr wrote:
> >> Are there any quirks in the support for Armenian? If there are
> >> please
> >> document that somewhere so that we can point people to the right
> >> place.
>
> > Uwe, why did you not answer this question? I am doing that now, because
> > I think that the quirks should be known.
>
> I explained the quirks I saw.
I was missing at least the one with \aroff. I know that this solution is OK
for you, but it is a quirk, isnt it?
> > - The same happens if you change the document encoding to the new
> > armenian encoding. In both cases the user has no clue why that happens.
>
> What's the problem here? When I set the encoding manually I know what I am
> doing.
You know that, but the user who chooses armscii because it is one of the
first encodings in the list and confuses it with ascii does not know that.
And if he does not immediately do a preview it can be very confusing for
him which change destroyed his document.
> Using an Armenian encoding for english of course gives senseless
> output. That's the same as if I write an Hebrew text but choose latin2 as
> encoding.
Sure. But the latter is less likely to happen, and more important it will
cause an error message talking about encodings.
> > - The lyx2lyx part is useless, since the armenian language tags are not
> > removed from the body, and because of the encoding problem Uwe got. A
> > working lyx2lyx conversion would need to translate all armenian unicode
> > characters into the pseudo ascii coding used by armtex.
>
> I don't think so. Converting an Armenian ocument to LyX 1.4.x is not
> posible as Armenian is only supported by LyX 1.5. That's something the
> user can accept. Furthermore translating the characters to the nasty
> armtex transliteration is not a good solution as this leads only to
> confusions. So the lyx2lyx code is only useless for reverting to another
> LyX 1.5svn version.
??? What you wrote above means that the lyx2lyx code is always useless. It
will produce an invalid file for earlier 1.5 formats, and not produce any
file for 1.4 and earlier formats.
> I mean we cannot be compatible backward to every old LyX-version. Some
> features are new in a certain LyX version without backward compatibility.
> I mean we currently can also not export all non latin1 documents to LyX
> 1.4.x-format. To test this, create an English document and insert there a
> latin2-only character. Then try to convert the document to LyX 1.4.x -
> lyx2lyx fails. I discussed this two days ago with José.
>
> Btw. I think we currently don't have an Armenian user, so why would
> somebody have it working in LyX 1.4.x?
Then why do lyx2lyx stuff at all? IMHO it only makes sense if it results in
a file that can be handled somehow by 1.4. This is not impossible, but
indeed some work.
> > - The test for the \usepackage{armtex} output is not correct (inputenc
> > == "default" is not handled correctly,
>
> I don't understand. When you choose Armenian as document language with
> default encoding, armscii8 will be used as input encoding and armtex is
> loaded. Armtex is also loaded when you chose the armscii8 encoding
> manually.
Look at the documentation of the inputenc member of BufferParams, especially
at the value "default" (which is _not_ the default encoding determined by
the document language).
> > and neither the case when armenian is one of several used languages)
>
> This isn't true. Take my example file I committed and mark some characters
> as e.g. Afrikaans. Then your get
> \usepackage[latin9,armscii8]{inputenc}
>
> This is correct but armtex can't handle this case, so not our problem or
> bug.
What is missing in this case is loading of armtex. I know that it is
academic, because this does not work anyway, but it would at least mark the
place in the code for future adjustments.
> > - The FORMAT entry is useless
>
> Feel free to make it better.
I am not going to do your homework. Look at the old entries, they always
explain the change in all details. This is important, since the FORMAT file
is basically the documentation of the file format.
> > - future implementation of proper armenian support is now much more
> > difficult, because existing documents with the ERT quirks need to be
> > handled.
>
> I disagree. As long as the armtex-package is not rewritten, it will stay
> like this. In the meantime the users can now use it with LyX. And later
> eventualy converting an ERT command via lyx2lyx shouldn't be difficult as
> there is only one, the \aroff command.
Exactly this lyxl2yx conversion will be very difficult to get right. I did
similar stuff (box<->minipage), the code is a mess and still not 100%
correct.
> To say it again, I think the current solution is useful also for average
> users and doesn't interfere
> with existing LyX documents. It furthermore can be adopted to any future
> solution, for example the
> adoption for the babel call is just to remove the antry in the languages
> file, the \aroff command can easily be handled as it is only one command.
I know that the current state is OK for you and I respect that. But since
Jose asked for quirks and you did only mention one I felt the need to list
the others. That does not mean that any of these needs to be fixed, but it
should at least be allowed to mention them.
Georg
