Re: Embedding fonts

On Sat, 8 Oct 2005, Gérard Talbot wrote:

I'm still interested to see whether this WEFT thingy is capable of
doing the *right* thing

My testings suggest WEFT - or rather MSIE 6 - does not do the right thing.

http://www.gtalbot.org/DHTMLSection/...nacomDemo.html

Thank you for taking the trouble to try this. Indeed we both seem to
come to the conclusion that it isn't working. But I don't know how to
discern whether it's the WEFT3 tool or MSIE which is the problem,
since there seem to be no detailed diagnostics for what's been put
into the downloadable font.

However, I do have a further question for you. When I try to display
your page (URL above) on MSIE6, I get empty rectangular boxes, which
is the same as what I get when no downloadable font is specified, and
no appropriate font is installed/available for MSIE to use.

However, when I try my own version of this, I get rectangular boxes
with diagonal crosses through them (as I mentioned before). The
diagonal crosses go away if I comment-out the specification of the
downloadable font (@font-face stanza specifying the .eot file) in the
CSS.

I note that the MS WEFT documentation talks about permissible URLs
from which the downloadable font can be used. Maybe the presence or
absence of these diagonal crosses is indicative of an error in that
area? I *thought* I had specified the permissible URLs correctly when
I made the downloadable font, but maybe not - I'm new to this WEFT
tool.

If you happened to know, one way or the other, what these diagonal
crosses mean, it would be helpful. If it seems useful (and if you
would consent to a lightly-modified version of your own HTML file
appearing in my test area, at least while we discuss it), I could try
to put this up on our web server.


[...]

Incidentally, on wider reading I see that recent versions of Netscape
browser can also do embedded fonts of some kind,

Only NS 4.7x I believe. Font synthesis .pfr (meaning Portable Font Resource)
are not compatible with MSIE.

Maybe it was a mis-reading of the page which I stumbled across. And
which I'm afraid I didn't bookmark and don't seem able to find again
now (it's not in my history stack either). Sorry! I really should
remember to bookmark stuff.

best regards

On Mon, 10 Oct 2005, Gérard Talbot wrote:

Alan J. Flavell a écrit :
[...]

Set this:

Tools/Internet Options.../Security tab/Internet web content zone
icon/Custom level/Downloads section/Font download/Prompt

Matter of fact, on the office machine where I had this working, that
setting is imposed by non-optional policy anyway.

I see rectangular boxes when I click Yes but I see all the correct glyphs when
I click No because nunacom is [permanently] installed on my system.

Thanks - that's useful input, I think.

However, when I try my own version of this, I get rectangular
boxes with diagonal crosses through them (as I mentioned before).
The diagonal crosses go away if I comment-out the specification of
the downloadable font (@font-face stanza specifying the .eot file)
in the CSS.

I don't know. You need to indicate to WEFT where on the web (domain
name) and where on your computer (local drives) you will be using
the webfonts.

Indeed, I understood that much - but I wasn't sure if I was doing it
right :-{ But see other followup, which I'm just about to compose.

If it seems useful (and if you would consent to a
lightly-modified version of your own HTML file appearing in my
test area, at least while we discuss it), I could try to put this
up on our web server.

Well, just ask.

I just did ;-)

What do you want me to do/change exactly?

I don't want you to change anything. I just want you to say that you
won't pursue me for copyright infringement if a modified version of
your page appears (temporarily, at least) in the test area on our
server.

cheers

On Mon, 10 Oct 2005, Gérard Talbot wrote:

I tried to subset code2000's Canadian Aboriginal Syllabics in WEFT
but this range isn't listed. Some others are not either like
Ethiopic and Cherokee.

I see what you mean...

code2000 does not support the Canadian Aboriginal Syllabics,
Ethiopic and Cherokee (1200 to 167F).

Curious. According to SIL ViewGlyph, the font *does* contain
the characters. But according to the MS Font Properties Extension,
there's no indication that the font supports these writing systems
in so many words, although there are some cryptic indications:

Reserved for Unicode SubRanges Bit xx

for various values of xx. This notation was unfamiliar to me.
But the values 75, 76 and 77 are amongst those present (see below why
this is significant).

If I could have an 1400 to 167F unicode font installed, I might be
able to do with WEFT a real accessible and web standards webpage.

AbSans (Aboriginal Sans) *also* contains the characters, and *also*
for this font there's no mention in the MS Font Properties Extension
that it supports these writing systems, and *again* there's this
mention of "Reserved for Unicode SubRanges Bit xx", in this case for
values 76 and 77.

Google finds this URL as a match for "Reserved for Unicode SubRanges":
http://partners.adobe.com/public/dev...ex_os2.html#ur

Indeed, bit 75 is Ethiopic, 76 is Cherokee, and 77 is Unified Canadian
Aboriginal Syllabics.

So it appears that these fonts are correctly marked, but the
diagnostic tool that I'm using (MS Font Properties Extension) doesn't
know how to correctly report them.

And it further seems that WEFT3 doesn't know either, based on what you
are reporting, and on what I also saw for myself.

However, MSIE6 *does* understand them, seeing that it's quite happy
via its "Internet Options"> "Fonts" dialog to select a default font
for "Canadian Syllabics", and offer a selection list which contains
(in my case, provided that they are installed) precisely these two
fonts, no more and no less.

Seems to me that we need an updated version of WEFT before we could
progress this any further. (I bet this would have been quite easy
with open source, grumble...).

thanks

On Mon, 10 Oct 2005, Gérard Talbot wrote:

Without the nunacom font, the word we see/can read is "eu3DN4tsJ5"
and taht is what is in the page as ascii (basic latin).

Which, according to HTML standards, *is* what the page contains.
That was my point.

Changing the encoding from ascii to utf-8 *in no way* improves that
situation, since for characters 0-127 the two encodings are identical.

*With* the nunacom font (which I categorise as "faked"), one may see
something different, in fact one may get the visual impression which
the misguided author intended. However, I repeat that this does not
accord with the HTML specification - it's a fake.

Your postings have gradually moved towards agreeing with me on this,
without quite saying it in as many words.

cheers

On Tue, 11 Oct 2005, Alan J. Flavell wrote:

But according to the MS Font Properties Extension,
there's no indication that the font supports these writing systems
in so many words, although there are some cryptic indications:

Reserved for Unicode SubRanges Bit xx

for various values of xx. This notation was unfamiliar to me.
But the values 75, 76 and 77 are amongst those present

Google finds this URL as a match for "Reserved for Unicode SubRanges":
http://partners.adobe.com/public/dev...ex_os2.html#ur

Indeed, bit 75 is Ethiopic, 76 is Cherokee, and 77 is Unified Canadian
Aboriginal Syllabics.

See http://www.microsoft.com/typography/unicode/ulu.htm
http://www.microsoft.com/typography/unicode/cscp.htm
for older versions. Apparently, your (any?) version of "MS Font
Properties Extension" isn't up-to-date.

On Mon, 10 Oct 2005, Alan J. Flavell wrote:

http://www.google.com/search?q=eu3DN4tsJ5
Google indeed has indexed a strange ASCII word, rather than any

Or my favourite

http://www.google.com/search?q=pwiksqwn

because Google itself is the first hit.

Canadian Syllabics.[1]
[1] It might be that Google doesn't yet index Canadian Syllabics,

Why not?
http://google.com/search?q=ppewww.ph...unidata14.html

On Mon, 10 Oct 2005, Gérard Talbot wrote:

When I double-click on the filename, I get the Inuktitut alphabet at the
place of the basic latin ascii (32-128).

Can't you see the contradiction? The Inuktitut alphabet is obviously
NOT basic Latin.

http://www.gtalbot.org/GRAPHICS/PNG/...tinNunacom.png

Oh dear! A Fro^Henchman with his sick keyboard layout even
f^Hmessed up the positions of the ASCII digits! This is pure horreur!

On Tue, 11 Oct 2005, Andreas Prilop wrote:

On Mon, 10 Oct 2005, Alan J. Flavell wrote:

Canadian Syllabics.[1]
[1] It might be that Google doesn't yet index Canadian Syllabics,

Why not?

I pasted-in what I took to be a "word" from a properly-encoded web
page (one that I had made myself locally), and Google found no matches
for the word.

You can now use this page from Gérard Talbot -
http://www.gtalbot.org/DHTMLSection/...nacomDemo.html - as an
online example to use in your search. Again I find no matches for any
of the words, leading me to conclude that Google isn't indexing them.
Also, Canadian Syllabics is not one of the writing systems mentioned
on their languages page, nor are the associated individual languages,
not even on the Canadian Google page:
http://www.google.ca/language_tools

http://google.com/search?q=ppewww.ph...unidata14.html

I think that proves that Google has indexed a (primarily Latin-1) web
page which happens to contain some of the relevant characters - but
that wasn't what I was investigating. A search for those characters
themselves, nor words built with them, does not produce any web pages.
"Unless you know better".

On Tue, 11 Oct 2005, Alan J. Flavell wrote:

You can now use this page from Gérard Talbot -
http://www.gtalbot.org/DHTMLSection/...nacomDemo.html - as an
online example to use in your search. Again I find no matches for any
of the words, leading me to conclude that Google isn't indexing them.

It takes some time before a (new) page gets into the index.
http://google.com/search?q=www.gtalb...nacomDemo.html

A search for those characters
themselves, nor words built with them, does not produce any web pages.

I don't know any other pages with *words* of such characters.
So let's wait until the above page is in Google's index.
(I have submitted it to Google.)

On Tue, 11 Oct 2005, Andreas Prilop wrote:

It takes some time before a (new) page gets into the index.

Sure! I was hoping that Google would find this word somewhere else.

I don't know any other pages with *words* of such characters.
So let's wait until the above page is in Google's index.

Here's a search string which will find you some pages already known to
Google, and which contain some properly-formed Can.Syl strings:

"Please download a Languagegeek.com font to view"

But when asked to search for those Can.Syl strings, Google comes up
empty handed. I can't see any other explanation than to assume that
Google isn't indexing these strings. (Yet, anyway).