Apropos emacs-intl-fonts & https://wiki.debian.org/DebianHebrew

Apropos emacs-intl-fonts & https://wiki.debian.org/DebianHebrew

[Jambunathan K]

I am trying to understand a report that mixes Hebrew and English in an
Org table and exports that Org file to ODT.  (The report concerns ODT
exporter and not Emacs).

This post is in two parts.

----------------------------------------------------------------
For understanding the issue, I installed culmus and Emacs is displaying
the Hebrew glyphs just fine.

In the following wiki page,

    https://wiki.debian.org/DebianHebrew

I also happened to run across a remark marked as 

    Do we need these?
    ....

    emacs-intl-fonts - Fonts to allow multi-lingual PostScript printing
    from Emacs 

    ....

Can someone knowledgeable fix the wiki page to clarify whether or not
emacs-intl-fonts is required or not for hebrew support?

----------------------------------------------------------------

[This for my basic Hebrew-101 or Bidi-101]

In the following message,

    http://lists.gnu.org/archive/html/emacs-orgmode/2013-10/msg01018.html

there is the following Org attachment, 

   http://lists.gnu.org/archive/html/emacs-orgmode/2013-10/txtgwNqsktCoy.txt

How would I make the English row correspond to it's Hebrew row.

----------------------------------------------------------------

[1] http://lists.gnu.org/archive/html/emacs-orgmode/2013-10/msg00719.html

Re: Apropos emacs-intl-fonts & https://wiki.debian.org/DebianHebrew

[Eli Zaretskii]

> From: Jambunathan K <kjambunathan@gmail.com>
> Date: Sun, 03 Nov 2013 13:10:50 +0530
> Cc: Uwe Brauer <oub@mat.ucm.es>
> 
> 
> In the following wiki page,
> 
>     https://wiki.debian.org/DebianHebrew
> 
> I also happened to run across a remark marked as 
> 
>     Do we need these?
>     ....
> 
>     emacs-intl-fonts - Fonts to allow multi-lingual PostScript printing
>     from Emacs 

That page was last updated in 2009, so it is terribly outdated.
Certainly, anything related to Emacs that it says is incorrect, since
Emacs 24.1 with bidirectional editing support was released in 2012.

> Can someone knowledgeable fix the wiki page to clarify whether or not
> emacs-intl-fonts is required or not for hebrew support?

They are not needed.  Any font that supports the corresponding
characters will do.

> In the following message,
> 
>     http://lists.gnu.org/archive/html/emacs-orgmode/2013-10/msg01018.html
> 
> there is the following Org attachment, 
> 
>    http://lists.gnu.org/archive/html/emacs-orgmode/2013-10/txtgwNqsktCoy.txt
> 
> How would I make the English row correspond to it's Hebrew row.

You mean, like this:

| שלום  | דוד   |
| ‏Hello | David |

The reason for this problem and the solutions available in Emacs are
described in the node "Bidirectional Display" in the ELisp manual.
Of course, what it says might not help at all with export to ODT.

Re: Apropos emacs-intl-fonts & https://wiki.debian.org/DebianHebrew

[Jambunathan K]

[-- Attachment #1: Type: text/plain, Size: 1119 bytes --]


Thanks for reading through the wiki page.

Eli Zaretskii <eliz@gnu.org> writes:

> You mean, like this:
>
> | שלום  | דוד   |
> | ‏Hello | David |

Not that.

What I see differs from what Uwe said he wanted.  See the attached
screenshots.  The "other" screenshot is from LibreOffice.  I cannot read
Hebrew so I have no idea what is Hello or what is David.

> The reason for this problem and the solutions available in Emacs are
> described in the node "Bidirectional Display" in the ELisp manual.

Warning: Newbie questions ahead.

I am just trying to get a quick help on Bidi.  I am trying to figure out
what markers would be needed to achieve the rendering in Emacs.  I can
use that as a baseline to form a basic understanding of Bidi.

Is there a Bidirectional mode (think Whitespace mode) whereby I can get
a visual feedback on the markers used in the text?

(I have consulted the Info node briefly. May be multiple readings would
be required to form a good understanding.)

> Of course, what it says might not help at all with export to ODT.

True. I understand that.


[-- Attachment #2: Bidi-What-Is-Desired.png --]
[-- Type: image/png, Size: 3062 bytes --]

[-- Attachment #3: Bidi-What-I-See-In-Gnus.png --]
[-- Type: image/png, Size: 6954 bytes --]

Re: Apropos emacs-intl-fonts & https://wiki.debian.org/DebianHebrew

[Uwe Brauer]

[-- Attachment #1: Type: text/plain, Size: 1347 bytes --]

>> "Jambunathan" == Jambunathan K <Jambunathan> writes:

   > Thanks for reading through the wiki page.

   > Eli Zaretskii <eliz@gnu.org> writes:

   >> You mean, like this:
   >> 
   >> | שלום  | דוד   |
   >> | ‏Hello | David |

   > Not that.

   > What I see differs from what Uwe said he wanted.  See the attached
   > screenshots.  The "other" screenshot is from LibreOffice.  I cannot read
   > Hebrew so I have no idea what is Hello or what is David.


May be I should try to clarify what I meant. (For some bizarre reason I
can't make gnu emacs 24 work with my gnus setting and that is why I have
to use Xemacs 21.5, which has no bidi support, I hope that the hebrew
text goes through without trouble.

What I have in mind is a interline translation.

So first comes a line in Hebrew and then the next line comes in English,
where however the words are displayed l2r, but the order of the words is
r2l.

This can be achieved in OO.

It can be achieved in org mode, without using tables.
Writing this, and exporting it to odt
שלום דוד.
david hello 


Works nicely.

Writing the same in tables and exporting



| שלום  | דוד   |
| david | hello |
 
Does not work, since the in the first hebrew line, the letters are
correctly displayed r2l, but the words are displayed l2r.


Uwe Brauer 

[-- Attachment #2: smime.p7s --]
[-- Type: application/pkcs7-signature, Size: 5556 bytes --]

Re: Apropos emacs-intl-fonts & https://wiki.debian.org/DebianHebrew

[Jambunathan K]

The hebrew word that reads like English "Tit", that is "David" right?

You want the Hebrew table row be read in R2L manner and the English row
down below correspond one to one with row above it.  (So the English
cells would have a directional override.)

Btw, the text that you have in Org tables, do they have directional
markers anywhere.  Or you just type the English text in "reverse" order
to get the desired effort.

Eli,

How would I quickly inspect directional markers in a text.  Should I
raise for a feature request for it.  Moving the cursor and waiting for
the cursor shape to change is just a pain.

Uwe Brauer <oub@mat.ucm.es> writes:

>>> "Jambunathan" == Jambunathan K <Jambunathan> writes:
>
>    > Thanks for reading through the wiki page.
>
>    > Eli Zaretskii <eliz@gnu.org> writes:
>
>    >> You mean, like this:
>    >> 
>    >> | שלום  | דוד   |
>    >> | ‏Hello | David |
>
>    > Not that.
>
>    > What I see differs from what Uwe said he wanted.  See the attached
>    > screenshots.  The "other" screenshot is from LibreOffice.  I cannot read
>    > Hebrew so I have no idea what is Hello or what is David.
>
>
> May be I should try to clarify what I meant. (For some bizarre reason I
> can't make gnu emacs 24 work with my gnus setting and that is why I have
> to use Xemacs 21.5, which has no bidi support, I hope that the hebrew
> text goes through without trouble.
>
> What I have in mind is a interline translation.
>
> So first comes a line in Hebrew and then the next line comes in English,
> where however the words are displayed l2r, but the order of the words is
> r2l.
>
> This can be achieved in OO.
>
> It can be achieved in org mode, without using tables.
> Writing this, and exporting it to odt
> שלום דוד.
> david hello 
>
>
> Works nicely.
>
> Writing the same in tables and exporting
>
>
>
> | שלום  | דוד   |
> | david | hello |
>  
> Does not work, since the in the first hebrew line, the letters are
> correctly displayed r2l, but the words are displayed l2r.
>
>
> Uwe Brauer 

Re: Apropos emacs-intl-fonts & https://wiki.debian.org/DebianHebrew

[Eli Zaretskii]

> From: Jambunathan K <kjambunathan@gmail.com>
> Date: Mon, 04 Nov 2013 01:50:38 +0530
> Cc: emacs-devel@gnu.org
> 
> The hebrew word that reads like English "Tit", that is "David" right?

Yes.

> You want the Hebrew table row be read in R2L manner and the English row
> down below correspond one to one with row above it.  (So the English
> cells would have a directional override.)

No, that won't work.  Directional override changes the directional
properties of individual _characters_, so "Hello" will be displayed as
"olleH", which is not what you want.

What you want is to cause _words_ be laid out right to left, even
though normally 2 juxtaposed English words would be laid out left to
right, in the direction they are read.  For that, use the RLM
(RIGHT-TO-LEFT MARK) between "Hello" and "David".  That's one solution
suggested by the ELisp manual.

> How would I quickly inspect directional markers in a text.

Why would you need to?  These control characters are supposed to be
invisible (the Unicode Standard actually advises to remove them from
text on display, but it was un-Emacsy, so I left them in).

I suggest to tell what problem are you trying to solve by displaying
the directional controls, rather than asking how to display them.
(And I already answered that question anyway: customize
glyphless-char-display-control.)

Re: Apropos emacs-intl-fonts & https://wiki.debian.org/DebianHebrew

[Eli Zaretskii]

> From: Uwe Brauer <oub@mat.ucm.es>
> Date: Sun, 03 Nov 2013 21:03:00 +0100
> 
> May be I should try to clarify what I meant.

I understand what you want.  In my response, I've shown how to display
what you want as you want it, using one of the methods described in
the ELisp manual.  There are other methods, described in the same
place, and I hope at least one of them will allow exporting stuff to
odt (whatever that means) and displaying it as you want in OpenOffice
(assuming the latter itself doesn't have bugs with bidirectional
display).

> Writing the same in tables and exporting
> 
> 

> | שלום  | דוד   |
> | david | hello |

> Does not work, since the in the first hebrew line, the letters are
> correctly displayed r2l, but the words are displayed l2r.

I repeat here my response:


| שלום       | דוד  ‏ |
| Hello      | ‏David |

As you see, words are displayed right-to-left in the second line as
well.

Re: Apropos emacs-intl-fonts & https://wiki.debian.org/DebianHebrew

[Jambunathan K]

Eli Zaretskii <eliz@gnu.org> writes:

> odt (whatever that means)

It means OpenDocument Text.  (OpenDocument Format that gives you a text
document)

A presentation file would be .odp => OpenDocument Presentation etc.

----------------------------------------------------------------

Btw, speaking of "Hello David" table, I though the solution would
involve a directional override whereby English words are forced to
behave as though they are Hebrew words.  In essence, there could be
multiple ways of getting the final effect and I would like to atleast
experiment and see the nature of markers affects the way LibreOffice
takes it hints.

----------------------------------------------------------------

I will revert on this thread after more soaking.  Thanks for the hint so
far, they give a good starting point.

Re: Apropos emacs-intl-fonts & https://wiki.debian.org/DebianHebrew

[Eli Zaretskii]

> From: Jambunathan K <kjambunathan@gmail.com>
> Cc: Uwe Brauer <oub@mat.ucm.es>,  emacs-devel@gnu.org
> Date: Mon, 04 Nov 2013 02:14:55 +0530
> 
> Btw, speaking of "Hello David" table, I though the solution would
> involve a directional override whereby English words are forced to
> behave as though they are Hebrew words.

Again, a directional override causes English _characters_ to behave as
if they were Arabic or Hebrew characters.  That is not what you want
in this case.

The ELisp manual describes the methods to achieve what you want:
affect the relative layout of words, without changing how each word is
displayed by itself.

Re: Apropos emacs-intl-fonts & https://wiki.debian.org/DebianHebrew

[Eli Zaretskii]

> From: Jambunathan K <kjambunathan@gmail.com>
> Cc: oub@mat.ucm.es,  emacs-devel@gnu.org
> Date: Mon, 04 Nov 2013 00:21:00 +0530
> 
> What I see differs from what Uwe said he wanted.  See the attached
> screenshots.  The "other" screenshot is from LibreOffice.  I cannot read
> Hebrew so I have no idea what is Hello or what is David.

Then I guess you want this:

| שלום       | דוד  ‏ |
| Hello      | ‏David |

It doesn't really matter, the solution (one of them) was demonstrated
by my previous example as well.

> I am just trying to get a quick help on Bidi.  I am trying to figure out
> what markers would be needed to achieve the rendering in Emacs.  I can
> use that as a baseline to form a basic understanding of Bidi.

If you really want basic understanding of bidi, you will need to read
the UAX#9, the annex to the Unicode standard that describes the
Unicode Bidirectional Algorithm.  That is what Emacs implements.

But IMO, that's overkill.  The node in the ELisp manual that I pointed
to includes a cookbook that you should be able to use to get this
stuff right, without understanding too much how it works internally.

> Is there a Bidirectional mode (think Whitespace mode) whereby I can get
> a visual feedback on the markers used in the text?

Not sure what you mean.  The bidirectional control characters by
default are visible (as thin spaces), and you can type "C-x =" on them
(or "C-u C-x =", if you want more details) to see which one is which.
Alternatively, you can customize glyphless-char-display-control to
display these characters as boxes with the Unicode codepoint spelled
out inside.

So the visual feedback is immediate without any special modes.  But
I'm not sure how this will help you: the effect of these marks on the
surrounding text is complex and cannot be predicted without a deep
understanding of the reordering algorithm.  You are welcome to study
it, but that is not needed just to solve your immediate problems.

> (I have consulted the Info node briefly. May be multiple readings would
> be required to form a good understanding.)

The manual gives precise instructions how to use each method, just try
them one by one and see which one fits best.

Re: Apropos emacs-intl-fonts & https://wiki.debian.org/DebianHebrew

[Jambunathan K]

Eli Zaretskii <eliz@gnu.org> writes:

>> Is there a Bidirectional mode (think Whitespace mode) whereby I can get
>> a visual feedback on the markers used in the text?
>
> Not sure what you mean.  The bidirectional control characters by
> default are visible (as thin spaces)

These thin spaces are not visible (unless you expect them to be there
and move the cursor over them).  In the example that you cited, there
are two RLM markers that become prominent only after I install the
customization down below.

> Alternatively, you can customize glyphless-char-display-control to
> display these characters as boxes with the Unicode codepoint spelled
> out inside.

This helps.

    (custom-set-variables
     '(glyphless-char-display-control
       (quote ((format-control . acronym)
               (no-font . hex-code)))))