Discussion of Emacs support for multi-directional text

Tim Freedom | 16 Jun 00:12

Picon

Bidi - will it happen ?

On Tue, 11 Jun 2002 08:39:23 +0300 (IDT),
  Eli Zaretskii eliz <at> is.elta.co.il wrote:
> 
> On Mon, 10 Jun 2002, Moe Elzubeir wrote:
> > So I take it we are at square 0 ?
> 
> Not exactly, but close.
> 
> > I mean, where do I look to see exactly
> > what needs to be implemented and what the agreed upon
> > approach(es) if any.
> 
> You can look in the archives of this mailing list.

I'm at a loss here, why is it that qemacs

  http://fabrice.bellard.free.fr/qemacs

has been able to include bidi where-as emacs hasn't ?
Mind you I haven't been able to compile the package
(to verify for myself), but his claims still stand.

I know nothing of lisp (let alone elisp), but can't
the inclusion of fribidi (fribidi.sf.net) be
incorporated somehow into the emacs code even as an
external call for the time being (irrespective of
inefficiency, ugliness, etc - given the user opts for
its inclusion) ?

There is a major disconnect on this topic if one were

(Continue reading)

Tzafrir Cohen | 16 Jun 00:53

Picon

Re: Bidi - will it happen ?

On Sat, 15 Jun 2002, Tim Freedom wrote:

> On Tue, 11 Jun 2002 08:39:23 +0300 (IDT),
> Eli Zaretskii eliz <at> is.elta.co.il wrote:
> >
> > On Mon, 10 Jun 2002, Moe Elzubeir wrote:
> > > So I take it we are at square 0 ?
> >
> > Not exactly, but close.
> >
> > > I mean, where do I look to see exactly
> > > what needs to be implemented and what the agreed upon
> > > approach(es) if any.
> >
> > You can look in the archives of this mailing list.
>
> I'm at a loss here, why is it that qemacs
>
> http://fabrice.bellard.free.fr/qemacs
>
> has been able to includebidi where-as emacs hasn't ?
> Mind you I haven't been able to compile the package
> (to verify for myself), but his claims still stand.

qemacs is not emacs.

qemacs is an editor with a number of similarities to emacs (mainly in the
looks and behaviour. However, it is more of a light-weight text editor.
Not much of a programable text-editor-and-a-kitchen-sink like emacs.

(Continue reading)

Eli Zaretskii | 16 Jun 08:09

Picon

Re: Bidi - will it happen ?

On Sun, 16 Jun 2002, Tzafrir Cohen wrote:

> fribidi is a library that mainly provides a method for converting logical
> Hebrew text to visual Hebrew text (in conformance with the unicode
> standard, and efficiently).
> 
> Actually, from what I understand, a similar algorithm has already been
> implemented in elisp by Eli.

Not exactly.  Handa-san wrote a batch-style Lisp implementation of bidi 
reordering, and demonstrated it 3 years ago.  However, Gerd Moellmann, 
who write the current Emacs display engine, was very unhappy with that 
implementation, and asked for something that doesn't work on large 
batches of characters (for fear that it will slow donw redisplay to an 
unbearable speed).

So I wrote a prototype of a sequential implementation of UAX#9, in C, 
which could be plugged directly into the Emacs display code.  The next 
step is to take that C code and actually put it into Emacs, making the 
necessary changes in the redisplay code as you go.

> Using fribidi or a similar library probably means patching emacs itself.
> It also requires porting to various emacs ports. elisp is probably more
> portable, although has some penalty in performance.

I don't believe an efficient solution for bidi can be coded in Lisp, 
since the redisplay works on the C level.

Tim Freedom | 16 Jun 03:04

Picon

Re: Bidi - will it happen ?

--- Tzafrir Cohen <tzafrir <at> technion.ac.il> wrote:
> On Sat, 15 Jun 2002, Tim Freedom wrote:
> 
> > On Tue, 11 Jun 2002 08:39:23 +0300 (IDT),
> > Eli Zaretskii eliz <at> is.elta.co.il wrote:
> > >
> > > On Mon, 10 Jun 2002, Moe Elzubeir wrote:
> > > > So I take it we are at square 0 ?
> > >
> > > Not exactly, but close.
> > >
> > > > I mean, where do I look to see exactly
> > > > what needs to be implemented and what the agreed upon
> > > > approach(es) if any.
> > >
> > > You can look in the archives of this mailing list.
> >
> > I'm at a loss here, why is it that qemacs
> >
> > http://fabrice.bellard.free.fr/qemacs
> >
> > has been able to includebidi where-as emacs hasn't ?
> > Mind you I haven't been able to compile the package
> > (to verify for myself), but his claims still stand.
> 
> qemacs is not emacs.
> 
> qemacs is an editor with a number of similarities to emacs (mainly in the
> looks and behaviour. However, it is more of a light-weight text editor.
> Not much of a programable text-editor-and-a-kitchen-sink like emacs.

(Continue reading)

Alex Schroeder | 16 Jun 16:32

Re: Bidi - will it happen ?

As to documentation, I gathered all I used at the time here:

http://www.emacswiki.org/cgi-bin/wiki.pl?CategoryBiDi

Alex
--

-- 
http://www.electronicintifada.net/diaries/index.html
http://www.us-israel.org/jsource/US-Israel/hr2506c.html

Eli Zaretskii | 16 Jun 08:13

Picon

Re: Bidi - will it happen ?

On Sat, 15 Jun 2002, Tim Freedom wrote:

> OK, so I'm comparing apples to organges, but the point that could be made
> is that there is someone out there that is interested in bringing-forth the
> same functionality - he should be contacted, at a minimum, to see if he has
> interest in helping out, no ?

As I wrote elsewhere, the problem is not to discuss what and how to do, 
the problem is to sit down and do it.

People who want to start coding the bidi support are most welcome.

> According to the archives, Eli has spread himself a bit too thin (his words,
> not mine).  What documentation is out there and what's the status of the
> implementation ?  What's left to be implemented in order to garner someone
> else's help in hopes of bringing forth a speedier solution until such time
> as when Eli gets freed-up ?

I believe I started to answer these questions in my other mail.  As for 
the documentation, there's nothing but the extensive comments in the code 
and UAX#9, but I believe that anyone who's familiar with what Unicode 
says about this should have no difficulty understanding the code I wrote; 
and my help is available.

> Eli, what do you envision doing to speed-up this process ?

Find more victims^H^H^H^H^H^Holunteers who would be ready to start 
coding, picking up where I left off.  For that, they will need to study 
the Emacs display code first, like I did at the time.

(Continue reading)

Eli Zaretskii | 16 Jun 08:04

Picon

Re: Bidi - will it happen ?

On Sat, 15 Jun 2002, Tim Freedom wrote:

> I'm at a loss here, why is it that qemacs
> 
>   http://fabrice.bellard.free.fr/qemacs
> 
> has been able to include bidi where-as emacs hasn't ?

There's nothing in principle that should prevent Emacs from supporting 
bidirectional editing, except that no one has stepped forward and 
volunteered to do the job.  I'm trying to do that by default, but my 
resources are in short supply lately.

> I know nothing of lisp (let alone elisp), but can't
> the inclusion of fribidi (fribidi.sf.net) be
> incorporated somehow into the emacs code even as an
> external call for the time being (irrespective of
> inefficiency, ugliness, etc - given the user opts for
> its inclusion) ?

We cannot use fribidi (or any other library that requires Emacs to pass 
it a chunk of text to be displayed) because that is incompatible with how 
the Emacsd display engine works.  I explained that several times here; 
you should be able to find the explanation in the archives.

> There is a major disconnect on this topic if one were
> to do any research on it (emacs+bidi that this) on
> google or anywhere else.  m17n.org had claimed that 
> bidi is close at hand (through its various symposiums).

(Continue reading)

Tim Freedom | 17 Jun 22:05

Picon

Re: Bidi - will it happen ?

On Sun, 16 Jun 2002 09:04:53 +0300 (IDT),
  "Eli Zaretskii" <eliz <at> is.elta.co.il> wrote:
> 
> On Sat, 15 Jun 2002, Tim Freedom wrote:
> 
[...]
>
> > There is a major disconnect on this topic if one were
> > to do any research on it (emacs+bidi that this) on
> > google or anywhere else.  m17n.org had claimed that 
> > bidi is close at hand (through its various symposiums).
> > Can't the various interests (m17n, emacs, xemacs, utf2000,
> > etc) unite and have one highly concentrated effort to get
> > this done with ?  Bidi's status is very nebulous at best.
> 
> The principles of bidi support and the general idea how to do it are 
> clear.  What is needed is motivated individuals who wouldn't hesitate
> to study the Emacs display engine code, then hack it to include the 
> bidirectional support.  I'll be more than happy to help such individuals 
> and pass to them what I've written so far; but first I need to see them 
> volunteer and start working seriously.
> 
> In other words, the problem is not the design; the problem is coding 
> (lots of it).  And since the first phase of that coding is to modify the 
> display engine, a person who wants to work on this will have to get a 
> fairly good understanding of how the display works in Emacs.
> 
> > Can we start anew with clear indications of what has
> > or hasn't been done and what is needed to complete the
> > job

(Continue reading)

Alex Schroeder | 17 Jun 23:26

Re: Bidi - will it happen ?

Tim Freedom <tim_freedom <at> yahoo.com> writes:

> With the various Bidi efforts out there and resident expertise, I don't
> think this issue would fall through the cracks given enough publicity
> about this topic and its needs.  Certainly Handa-san, et al might want
> to re-jump in, no ? 

Well, I think all the probable volunteers already know about it,
because it is mentioned from time to time on emacs-devel, for example.

Personally, I would really like the code available for all to see,
download, etc., but since I did not write it, I can not demand it.  :)
Eli has been very helpful in the past, however, so if *you* volunteer,
or bring a volunteer to this list, who will actually do some coding,
then I think you will find all the help you might need.  The point is,
Eli doesn't have the time, I guess, to explain this to people who will
probably not do the coding anyway.  Yes, it is an additional hurdle
for potential newcomers, but I think it is understandable given the
current constraints...  :(

Alex.
--

-- 
http://www.electronicintifada.net/diaries/index.html
http://www.us-israel.org/jsource/US-Israel/hr2506c.html

Eli Zaretskii | 18 Jun 07:34

Picon

Re: Bidi - will it happen ?

On Mon, 17 Jun 2002, Alex Schroeder wrote:

> Well, I think all the probable volunteers already know about it,
> because it is mentioned from time to time on emacs-devel, for example.

I'd expect that as well.  Also, emacs-bidi is a public mailing list, so 
anyone who is interested can easily subscribe and read what's available.

> Personally, I would really like the code available for all to see,
> download, etc., but since I did not write it, I can not demand it.  :)

I've told in the past that I don't want to publish unfinished code with 
my name all over it.  However, if someone tells me that he/she is 
prepared to actually start working on this, I promise I'll send the code 
and every auxiliary file/info I have that could help.

> The point is,
> Eli doesn't have the time, I guess, to explain this to people who will
> probably not do the coding anyway.

That's true, thanks for the understanding.  Yes, my free time is severely 
limited lately, and will remain limited for the months to come.  I might 
have some break in mid-August and around September, due to public 
holidays, but I'd rather sit and code when I have that time than posting 
more explanations or working on a site dedicated to the issue.

Having said that, I will find time to help, both by discussion and code, 
to anyone who shows a real intent to work on coding the bidi support.

(Continue reading)

Tim Freedom | 20 Jun 03:17

Picon

Re: Bidi - will it happen ?

--- Eli Zaretskii <eliz <at> is.elta.co.il> wrote:
> 
> On Mon, 17 Jun 2002, Alex Schroeder wrote:
> 
> > Well, I think all the probable volunteers already know about it,
> > because it is mentioned from time to time on emacs-devel, for example.
> 
> I'd expect that as well.  Also, emacs-bidi is a public mailing list, so 
> anyone who is interested can easily subscribe and read what's available.

With all due respect, I did a search on google and a few other search
engines in hopes of finding how would a volunteer know/hear of this
forgoing private email.  I did a search on "emacs-devel" and "bidi"
and got virtually nothing (the only detailish mention of emacs and bidi
including status is on this list, for what its worth).  I didn't find
any search abilities specific to either list and so I might be a bit
naive in my thinking, but from my experience no one who hasn't been
previously involved or exposed to bidi in the past would know what bidi
is and/or its requirements.  Furthermore, if the info is presented such
that no Bidi knowledge is required to complete the code, I would guess
a greater deal more people would want to look into this as it would fit
within their knowledge base (ie. they don't have to learn something new).

I went back to emacs-devel (raw text search) for the past 7 months and
saw no mention of Bidi status/help-call.  I'm not refuting your expert
expectations, but what's the harm in making explicit mention of Bidi's
current status and what's required to complete it (again 2-3 sentences
and maybe point at URLs of the posts on where this is discussed) on
emacs-devel and comp.emacs.  At a minimum it would further raise
awareness.

(Continue reading)

Eli Zaretskii | 20 Jun 08:06

Picon

Re: Bidi - will it happen ?

On Wed, 19 Jun 2002, Tim Freedom wrote:

> > I'd expect that as well.  Also, emacs-bidi is a public mailing list, so 
> > anyone who is interested can easily subscribe and read what's available.
> 
> With all due respect, I did a search on google and a few other search
> engines in hopes of finding how would a volunteer know/hear of this
> forgoing private email.  I did a search on "emacs-devel" and "bidi"
> and got virtually nothing (the only detailish mention of emacs and bidi
> including status is on this list, for what its worth).

Sorry, I don't understand: did your Google search find the emacs-bidi 
list or didn't it?  If it did, then that's all I'd expect: the next step 
is to study the archives of this list.

If Google did NOT find this list, then please tell what words did you use 
in Google.

> I didn't find any search abilities specific to either list

You can search with your Web browser.  Or you can download the archives 
to your local machine and search with whatever tools you have there.

> and so I might be a bit
> naive in my thinking, but from my experience no one who hasn't been
> previously involved or exposed to bidi in the past would know what bidi
> is and/or its requirements.

I don't see how someone who doesn't know what bidi is can help us add

(Continue reading)

Alex Schroeder | 22 Jun 20:01

Re: Bidi - will it happen ?

Eli Zaretskii <eliz <at> is.elta.co.il> writes:

>> I'm not refuting your expert
>> expectations, but what's the harm in making explicit mention of Bidi's
>> current status and what's required to complete it
>
> Nothing is wrong with that; I simply don't have time and resources to 
> create and maintain such a page.  If someone does that, I'll applaud.

All I know and learned and read I added to the CategoryBidi page on
www.emacswiki.org.  I even added reports about the display of Arabic,
so I think I did what is possible without actually doing the coding
itself.  :)  And since this is a wiki, anybody (not Eli, all the
lurkers on the list) can improve the pages.

Alex.
--

-- 
http://www.electronicintifada.net/diaries/index.html
http://www.us-israel.org/jsource/US-Israel/hr2506c.html

Eli Zaretskii | 23 Jun 06:43

Picon

Re: Bidi - will it happen ?


On Sat, 22 Jun 2002, Alex Schroeder wrote:

> All I know and learned and read I added to the CategoryBidi page on
> www.emacswiki.org.  I even added reports about the display of Arabic,
> so I think I did what is possible without actually doing the coding
> itself.  :)

That's great, thanks!

Yair Friedman (Jerusalem | 20 Jun 09:00

Re: Bidi - will it happen ?

On Wed, 19 Jun 2002 18:17:57 -0700 (PDT), 
Tim Freedom <tim_freedom <at> yahoo.com> writes:
> I went back to emacs-devel (raw text search) for the past 7 months and
> saw no mention of Bidi status/help-call.  I'm not refuting your expert
> expectations, but what's the harm in making explicit mention of Bidi's
> current status and what's required to complete it (again 2-3 sentences
> and maybe point at URLs of the posts on where this is discussed) on
> emacs-devel and comp.emacs.  At a minimum it would further raise
> awareness.
>

I think the emails Eli reference are these all from mid-January 2001. 

<Pine.SUN.3.91.1010118130031.5194A-100000 <at> is>
<y26ae8pqdvj.fsf <at> chaos.cs.huji.ac.il>
<Pine.SUN.3.91.1010118140124.5194F-100000 <at> is>
<200101190836.RAA07785 <at> etlken.etl.go.jp>
<86g0ifkgg6.fsf <at> gerd.segv.de>

There are other emails around that time that might add more light.

Return

Return to gmane.emacs.bidi.

Search Archive

Language

Change language

Options

Current view: Threads only / Showing only 20 lines / Not hiding cited text.
Change to All messages, whole messages, or hide cited text.

Post a message
NNTP Newsgroup
Classic Gmane web interface
XML

XML

RSS Feed
List Information

About Gmane