Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 1 Apr 2020 08:47:07 +0430
From:      Javad Kouhi <javad.kouhi@gmail.com>
To:        "Danilo G. Baio" <dbaio@freebsd.org>
Cc:        Hiroki Sato <hrs@freebsd.org>, freebsd-translators@freebsd.org
Subject:   Re: Self-introduction: Javad Kouhi
Message-ID:  <CA%2Bg814eqv9JAwsSP4OdY47L2f%2BkjQ=VNf%2B1oYUY2TGkBMjP%2BvA@mail.gmail.com>
In-Reply-To: <20200331204305.gyq2mlh4mv7lykob@t480.local>
References:  <CA%2Bg814e5=ipStdmvPFhPaDUprmnaOapcMb1xS9fcFNCA5Y9fQQ@mail.gmail.com> <20200331.124527.1078840241867193092.hrs@FreeBSD.org> <20200331170132.gyjlo7nel3cajidf@t480.local> <20200401.035204.1608271793023335727.hrs@FreeBSD.org> <20200331204305.gyq2mlh4mv7lykob@t480.local>

next in thread | previous in thread | raw e-mail | index | archive | help
Hello and thank you for the replies.

> Could I suggest starting with small documents, like
> articles/leap-seconds or articles/ipsec-must ?

I'll prepare a small article, so we would have something to start
with, will let you know when I'm finished.

> The UI is doing something because the string is saved differently in the
> repository:
> https://people.freebsd.org/~dbaio/weblate/2020-03-31-093415_747x699_scrot=
.png

The UI is showing the code in RTL format. In this form, the Persian
text is shown correctly, but the tags are not. On the other hand, in
this screenshot:

https://people.freebsd.org/~dbaio/weblate/2020-03-31-093415_747x699_scrot.p=
ng

The code is shown in LTR format.  In this form, the tags are shown
correctly, but the Persian text looks messy.

If one is shown correctly, the other would look messy. They can't be
shown correctly at the same time. This is a universal problem with RTL
languages and is not specific to Weblate, which in that case, we could
have simply do the translation somewhere else and commit them here
when they're ready. The problem is that, even if we take the source
code somewhere else like a desktop text editor, it will still suffer
from the same problem (I'm yet to see a document/text editor that can
handle it correctly). Sometimes it doesn't matter (as in most web
pages which are usually generated automatically), sometimes it does
matter, when people have to directly deal with the tags.

> I've built these initial translations you made, without any
> modification, see the results here:
> https://dbaio.bs2cloud.com.br/doc/fa01/fa_FA.UTF-8/books/handbook/

Although this initial build is based on just some test sentences, it
looks very promising. So, the compiled form seems to be already OK,
the main problem is with the source code.

> Does "translating tags" mean replacing the tag name of <foo> with
> Parisian counterpart in RTL, or the whole tag including <> in RTL?

I mean something like <=D9=84=DB=8C=D9=86=DA=A9>http://www.FreeBSD.org</=D9=
=84=DB=8C=D9=86=DA=A9> instead of
<link>http://www.FreeBSD.org</link>. This is the solution adopted by
similar projects, such as the Persian Wikipedia and the xepersian
package for LaTeX. I don't know if that's possible here, I guess it
requires fundamental modifications to the docproj infrastructures. I
could do the translation in the case those modifications to the
infrastructures are not a concern.

> Could you show a typical HTML file in UTF-8 Persian?  I would like to
> know what tags look like.

Is this good enough?
https://kouhi.org/sample.html

On Wed, Apr 1, 2020 at 1:13 AM Danilo G. Baio <dbaio@freebsd.org> wrote:
>
> On Wed, Apr 01, 2020 at 03:52:04AM +0900, Hiroki Sato wrote:
> > "Danilo G. Baio" <dbaio@FreeBSD.org> wrote
> >   in <20200331170132.gyjlo7nel3cajidf@t480.local>:
> >
> > db> >  Could you show a typical HTML file in UTF-8 Persian?  I would li=
ke to
> > db> >  know what tags look like.
> > db>
> > db> I would like to see some examples as well.
> > db>
> > db> Persian was already marked as RTL on Weblate.
> >
> >  Yes, but it still needs direction attributes in the DocBook results
> >  even if editors and input fields on the web browsers support RTL in
> >  the user interface.  We have options such as "RTL by default and
> >  marking up the LTR parts" and vice versa.  And RTL and LTR in the
> >  same #PCDATA with no tag will not be rendered correctly.  Probably we
> >  need to take care of them in the XSLTs after checking what "natural
> >  way" to write a marked-up document in HTML or XML is.
>
>
> Thank you both for the explanations.
>
> So, this will be an interesting issue to fix, let's check some real
> issues in our documents and the best way of fixing it.
>
> We will need your help Javad.
>
> Could I suggest starting with small documents, like
> articles/leap-seconds or articles/ipsec-must ?
>
>
> >
> > db> I've built these initial translations you made, without any
> > db> modification, see the results here:
> > db> https://dbaio.bs2cloud.com.br/doc/fa01/fa_FA.UTF-8/books/handbook/
> > db>
> > db> If this is good, I can go forward and add a Jenkins job to build it=
 regularly.
> >
> >  The LANG_TERRITORY part must be fa_IR, not fa_FA.
>
> I'll fix that, thank you.
>
> --
> Danilo G. Baio (dbaio)



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CA%2Bg814eqv9JAwsSP4OdY47L2f%2BkjQ=VNf%2B1oYUY2TGkBMjP%2BvA>