W3C home > Mailing lists > Public > w3c-wai-ig@w3.org > July to September 2000

[Fwd: Misconception?]

From: William Loughborough <love26@gorge.net>
Date: Thu, 03 Aug 2000 06:42:43 -0700
Message-ID: <398976D3.71786C0E@gorge.net>
To: ig <w3c-wai-ig@w3.org>
Chris Lilley is chair of the SVG Working Group so I asked his comment re
David Wooley's concerns about accessibility problems posed by SVG.


attached mail follows:

William Loughborough wrote:
> On the IG list David Wooley says:
> That means that you normally get individually placed characters, rather
> than whole words with a horzontal stretch factor, as allowed by the PDF
> primitives. 

> This can confuse tools that try to extract words from the
> text (I predict that the same problem will happen with SVG if, as I
> suspect, people use it for whole pages, not just graphics)."

It is certainly *possible* to use singly placed characters in SVG, but
there is no need to and significant disadvanteges to doing so. I have
rarely seen examples which do this, and the couple that I have have been
machine translations from PDF....

SVG allows the stylesheet to select a condensed or expanded font. It allows
the length of overall run of text to be specified, and the run then
adjusted at layout time to be that length. Hmm thats not clear, let me give
an example:

Suppose for example I lav out some graphics, which include text, in
Helvetica. Now thats a widely used font but there are lots of variations
from different font foundries and oin different platforms, plus Windows
thinks that Helvetica is spelled 'Arial'. So I specify in the SVG file the
assumed length of this text string, then if your Helvetica and my Helvetica
differ an adjustment is made so the line is the correct length. This
prevents, for example, the text being 5% too long and overwriting something
else, making it anywhere from untidy to illegible.

Other things - SVG requires UAs that support a clipboard to allow text to
be selected (yes, even if its all spaced out, vertical, on a curve,
whatever). Placing each letter as a separate text string messes this up, so
people don't do it. Its a;lwats helpful if an accessibility advantage is
tied to a non-accessibility advantage. In this case, screwing around
withthe text stops it being indexed correctly and thus means that less
people find your stuff with search engines.

The other occasion that I have sen unusual breaks in text is again with
automatically generated conversions from PostScript, PDF, Quark, etc where
the UA has done kerning and this is output as different text strings. Now
it doesn't have to be this way - SVG fonts include kern tables, so it is
better to keep entire strings together and use the kerning facility rather
than faking it with sliced up, positioned strings. It all depends on what
you are converting from and, as ever, how much information that source
format retained.

I converted a nice map example from PDF to SVG, and severalof the strongs
had kerning like this. because SVG is a textual format I was able to load
the file into a plain text editor and fix this up, joining the strings back

I don't see a particular connection between the use of SVG for entire pages
as opposed to single graphics, and the likelihood or not of text strings
being undesirably split.

> Could you please allay fears of this, lay this ghost, whatever. 

I hope that I have done so.

> If the
> rumor spreads that SVG is going to be the new PDF-style nemesis for
> accessibility, it can't help.


While we are o the subject of accessibility, I would like to raise a new
perspective. As some people are aware, i have a rather broad view of
accessibility, which includes for example the fact that many people do not
speak or read English (when asked what single thing would most improve
accessibility for most people, I replied a voice browser that spoke
Cantonese and Urdu).

So, it seems that many people agree that accessibility for the non-literate
is an issue. Lots of people speak English but cannot read it, for example.
And lots of people don't read English but can understand diagrams. Thus,
pictorial material can be am accessibility aid. If the thousand words of
explanation don't help because you can't read them, or because they are not
in a language you understand, then the picture may indeed be worth a
thousand words.

I would also point out that SVG has the ability to switch on the users
preferred language and conditionally display text (and graphics). So for
example it is easy to make a multilingual graphic that covers English,
Frennch, German, Spanish. This is likely to be only fractionally larger
than a monolingual one, but more accessible (according to my definition). 

The Jackaroo viewer, from the Koala team at INRIA, implements this
switching. It also implements user CSS stylesheets by the way, and has a
nice DOM viewer, so I encourage people to have a look at it.


Received on Thursday, 3 August 2000 09:43:40 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 13 October 2015 16:21:09 UTC