[WEB4LIB] Re: Arial MS Unicode Font

Andrew Cunningham andrewc at mail.vicnet.net.au
Wed Aug 21 21:14:06 EDT 2002


Hi all

Howard Pasternack wrote:
> A couple of comments about Arial MS Unicode. The great advantage of this 
> font is that it is in fact Unicode and that it supports a great variety of 
> glyphs in various Western and Asian languages.  With multi-character opacs 
> it works very well, especially when one record is in Korean, and a 
> scholarly work about the same topic might be in Japanese. The fonts 
> distributed with the Global IMEs do not provide the same functionality, and 
> in the case of opacs which output in Unicode, the fonts do not even display 
> all the glyphs correctly.  So, they really are not a 
> substitute.  Bitstream, has also stopped distributing its Unicode 
> font.  So, the only real alternatives are the commercial products.
> 

I'm very aware of the problem.

There are very few complete fonts, and the reality is that the glyph 
repetroies now required for a unicode font are beyond the current limits 
of the font etchnologies. For instance there are over 60,000 CJK 
characters alone now taht they've started populating the planes beyond 
the BMP.

Very few fonts vendors are willing to create complete unicode fonts. 
Probably cann't be done. The additional conjuncts and ligatures required 
for Indic languages for insatnce would be excessive. Code2000 and 
Code2001 are one of the few fonts still atttempting to keep abreast of 
the unicode standard.

Microsoft along with major font vendors seem to prefer fonts that 
implement a subset of unicode.

This is where we encounter problems. The older versions of netscape 
assume that one unicode font fits all. Which has never really been true.

IE on the other hand uses script subsets to assign default fonts. Added 
to this newer versions of the OS use font-linking technologies. If an 
appropraite font isn't available they'll determine which font to use.

Another problem is that none of the current browsers have implemented 
support for the :lang selector in css. Which would make life much much 
easier when langauge tagging was possible.

Although Bitsreams font is still available from companies like Netscape 
that licensed it. Although its still at version 2.0 and not being updated.

WRT to Arial Unicode MS, the font is not complete. There are lots of 
version floating around with different sets of glyphs. I've seen 
versions that include support for Hindi/Marathi/Tamil and versions that 
do not.

It has major problems esp with combining or stacking combining 
diacritics. MarkToBase and MarkToMark opentype features are not present 
in the font, maybe except for the Vietnamese subrange.

Arial Unicode MS is a commercial unicode font that microsoft shipped 
with some of its products. It was available for download for users that 
used the ublisher 2000 product, since it was not shipped on the cdrom. 
Installing the font on a computer that did not have an appropraite 
software or operating system installed was a licensing violation.

Verson ).84 was shipped with the Office 2000 software. This version was 
designed to support the Unicode 2.0 standard (quite old!).

> The issue is really not one of configuring an indiividual workstation in 
> the Library.  That is a fairly simple task.  The issue is assisting a user 
> outside of the library to configure a PC to read non-Western scripts output 
> the library.  Many opacs, my own included, provided a help screen which 
> directed the user to download the Microsoft font and to configure the 
> browser accordingly.  The instructions were simple and easy to follow for a 
> user with modest technical skills.  However, the links are now dead and a 
> substitute is not at hand.
> 

In theory you should have been directing them to read the licensing 
restrictions for the font and if they met the conditions then they could 
downlaod it.

The reality is that locating and installing fonts are always 
problematic. art of your problem is actualy how the web browsers have 
inplemented the ability to select default unicode fonts.

Part of the problem is that the OPAC has not been properly localised.

You only need complete fonts with the OPACs within your library. I'd 
assume that your users accessing the OPAC externally would only need a 
subset of unicode that is relevant to the languages they want to view. 
For full CJK support try Bitstream Cyber CJK 
[ftp://ftp.netscape.com/pub/communicator/extras/fonts/windows/CyberCJK.ZIP]

You also have Bitsream Cyberbit and Bitstream Cyberbase available there 
in the same directory. As mentioned earlier its version 2.0. And quite old.

> The problem will disappear eventually as more machines upgrade to XP or get 
> replaced.  But for the moment it is there.  My own institution will be 
> looking at site licensing NJStar.  -- Howard
> 

Assuming you're not using Netscape 4, a move to WinXP will solve your 
problems, including IMEs. Although with unicode support it is necessary 
to remember that IE and Netscape 6/7 use different font rendering 
technologies, so when you're dealing with complex script rendering there 
will be major diferences.

Not sure that the added expense of the NJStar products are required. But 
thats only my personal view.

Andrew




More information about the Web4lib mailing list