Re: lynx-dev Display of SGML Greek Math entities -- solution

[Steve White]

Henry, 
pardon me, in advance.
>
>> On terminals that can display Greek glyphs, the SGML Greek
>> Mathematical symbols should be rendered using those glyphs.  On
>> terminals that cannot display Greek glyphs, these entities must be
>> displayed distinctly from other characters.
>> 
>> Otherwise, information is lost.  This is not a matter of personal
>> preference, it is a matter of the purpose of the SGML Greek

>...(Although the hair on the back of my neck raises with "must be." :)  

We could build a browser that renders all characters as asterisks.
That would save a lot of screen real-estate, and relieve us of most 
concerns about information being preserved.  Another idea is:  To 
render all alphabetic characters as vowels.  Sometimes, meaning 
may be lost, but then it may be gained in Hawaiian or Hungarian.

This would also satsify those who are worried about their text-based
browser being slow.

>What I was referring to as a "personal preference" was your
>desire to spell out the entire name of the letter rather than consider
>some symbolic (perhaps two-character, perhaps three) representation of
>"Greek glyphs."  Using the name makes a width difference of 5 character
>cells (xi or pi versus omicron or upsilon -- even a greater difference
>if you want to include thetasym).  While this may not bother you, it may
>someone else.
>

I have made it clear elsewhere that it is my personal preference to 
represent them with a textual representations of their names.  This is 
because my experiments with math formulas using this representation 
worked out very nicely.  

Your argument about saving space is valid, though.

And who's responsible for thetasym, upsih, and piv, anyway?  Why not hbar?

>
>While it may seem as though I missed your point, I suspect likewise
>you missed mine.  Namely, in this day and age, a computer running an
>OS that _can't_ display Greek letters and a variety of other symbols
>is almost unheard of.  That same vintage NEC PC9801VX now runs NetBSD
>1.4.3_ALPHA, which supports a dozen or more locales.  I'm sure a current
>distribution (not available for the i386/pc98 platform) would support two
>or three times that number.
>
Henry, which of us couldn't run Mozilla?

>I still differ with you.  Lynx's behavior is not broken by the simple
>fact that anyone, as you yourself have done, may edit the translation
>tables to their satisfaction and re-compile.  In other words, if someone
>doesn't like what Lynx falls back to, they're free to change that.  My
>real point is, though, that there shouldn't be a need for anyone to fall
>back to some half-baked default.
>

Good grief.  Are you saying that open source software can't be broken,
because "anyone" can hack it?  Wow!  And Microsoft software can't be 
broken, because it is the standard!  

To think we ever concerned ourselves with quality!

Besides, I'm not even convinced I will be able to fix this, because the
Asian character translation is so labyrinthine.  I don't think "anybody" 
can fix this problem.

>> You aren't even looking at the problem in question.
>
>I'm sure that's true because I don't even see a problem (that can't be
>easily fixed).

We agree on something!

>
>> Your example doesn't contain any SGML Greek Mathematical symbols
>> such as "Α" but rather, explicit numerical references such
>> as "&945;" to characters in you document's character set.
>
>You want "Α" and "Α" to NOT represent the same thing, namely
>the capital letter Alpha?  Likewise, either "α" or "α" to
>NOT represent the same lower case alpha?  Why?  They're synonymous.
>I revised my Greek letter page to illustrate.

Different ranges of the HTML 4 SGML entities have been deliberately
chosen to have a tight relation with certain ranges of different
character sets, but unless you're using full Unicode, I don't think
you get them all.  ( I take this back...the latest releases of 
JIS may well contain all of the HTML 4 SGML character entities. 
Perhaps this explains your perplexity. )

The SGML 'Α' is not synonymous with the numeric character 
reference 'Α', Henry.  Generally, the numeric character references 
such as 'Α' refer to a character **in the document's character set**.
By contrast the SGML entity 'Α' is to be rendered as a mathematical
symbol Alpha, regardless of whether a corresponding character exists is 
in the document character set OR the display character set. 

For example, if the document character set is iso-8859-15, where
is '∴'?  Where is '∑'?

This point is often lost.  I even found a passage in the HTML 4 
document where it was lost.  But there are other places where it is 
made very explicit.

>(BTW, the page was not
>intended for this thread.  It normally resides, without the Lynx
>screenshots, on my lab server for student reference only.  I'll take it
>down in a couple of days.)
>

It has been very instructive.  I clearly still don't understand
something about how lynx processes EUC-JP internally.  I still don't
know if I can make a patch for this situation.  I'm not even sure
whether my patches work or not in multibyte character sets, because I'm 
pretty sure that my distribution version of lynx isn't working right 
for them yet.

>
>But really if there is anyone on this list who CANNOT produce Greek
>characters on their display, don't put up with it!  Get the fonts!  It
>has nothing to do with Lynx.
>

Say I have an 8-bit terminal.  I want to look at a document written
in Russian that contains math formulas containing Greek letters.

Here it's not a matter of fonts.  It's a matter of fitting all three 
alphabets into 8 bits.  Something has to give.

Are we to give up on lynx support for 8-bit displays?  That would
simplify its code a great deal!

Furthermore, I have spent the better part of a day trying to get
a terminal and lynx configured properly **Just To Look At Your Page**, 
and still I don't quite have it.  I bow to your abilities as a sysadmin.
Not Everybody has these Powers.  You are Special.

Or perhaps, this point is lost on you, because somebody else 
configured your system for you.

You are spoiled by being in a 2-byte EUC-JP environment, Henry.
Most of us don't live there yet.

Yours is a spurious argument, I'm afraid.   An argument of this form 
is often used to dismiss lynx altogether.  In fact, an argument of 
this form can be used to dimiss anybody who doesn't agree with me!

If you don't agree with me, just fix yourself!  (for example)



; To UNSUBSCRIBE: Send "unsubscribe lynx-dev" to address@hidden