Re: [Rswg] SVG/ASCII art and accesibility (was: Re: Fwd: I-D Action: draft-carpenter-rswg-rfc7996-issues-00.txt)

["Martin J. Dürst" <[e-mail]>]

Hello Jean, Alexis, Brian,

On 2022-10-09 01:21, Jean Mahoney wrote:
> Hi Brian,
> 
> On 10/7/22 6:17 PM, Brian E Carpenter wrote:
>> Jean,
>>
>> On 08-Oct-22 10:55, Jean Mahoney wrote:
>>
>> <snip>
>>
>>> [JM] Repeating here the comment that I've added to issue #20 
>>> (https://github.com/rfcseries-wg/new-topics/issues/20):
>>>
>>> The RPC is updating its procedure for checking artwork to include 
>>> reviewing the contents of the SVG <desc> and <title> elements and the 
>>> <artwork alt=""/> attribute (when xml2rfc supports it. Issue opened: 
>>> ietf-tools/xml2rfc#898 
>>> <https://github.com/ietf-tools/xml2rfc/issues/898>). If the artwork 
>>> lacks an accessible description, during AUTH48 the editor will ask 
>>> the authors to please provide text. We will also update the Online 
>>> Portion of the Style Guide to recommend to authors that they provide 
>>> descriptive text for artwork. rfc7322bis 
>>> <https://datatracker.ietf.org/doc/html/draft-flanagan-7322bis-07#section-3.7> already captures a recommendation regarding artwork accessibility.

Many thanks for this; I think this is the best way to proceed because it 
has the lowest overhead.

>> That's excellent, of course. But I still lack
>>
>> a) any drawing tool for SVG that supports <desc>;
> [JM] Ack. I've manually inserted <desc> into SVG embedded in RFCXML, and 
> it works (that is, the contents appear in the HTML output), but it would 
> be good to identify drawing tools that support <desc>.

Just a sample of one:
Inkscape would do that, under Object -> Object Properties -> 
Description. But then, Inkscape doesn't produce clean SVG.

Personally, I'd probably go with just adding the <desc>s by hand in the 
markup, which to me looks faster than selecting submenus over and over. 
But I understand that may not be for everybody.

>> b) any simple way of testing whether a <desc> or <artwork alt=""/> 
>> actually works with a screen reader in the html rendering;
> [JM] I'm also looking for a good screen reader or browser plugin to test 
> these things. Anyone have suggestions?

As far as I'm aware, for English at least, the commercial option for 
Windows is JAWS 
(https://www.freedomscientific.com/products/software/jaws/). On Windows, 
there is also a built-in screen reader called Narrator. The open source 
screen reader for Windows most widely used is NVDA. Mac has VoiceOver 
built in. For Android, it's TalkBack.

There are two problems to be aware of:
1) Screen readers can be expensive for the occasional try, because it's 
a small market and those who really need a screen reader (or their 
companies) are ready to pay a decent price.
2) Screen readers have a very steep learning curve. Interaction mostly 
happens with a large number of Ctrl/Alt/Fn/... key combinations. The 
default settings may not be the ones you prefer, and you may not know 
which settings you would prefer, anyway. A screen reader may start to 
talk, and be difficult to stop, or may stop, and be difficult to start 
again. It may be best to use two computers, one with the manual and 
another with the screen reader, because otherwise, the screen reader 
will start to read the manual whenever you want to check something in 
the manual.

https://weboverhauls.github.io/demos/svg/ shows tests for some relevant 
combinations of SVG elements, ARIA attributes, browsers, and screen 
readers, although it is quite a bit outdated.

>> c) a good example of a well annotated diagram in xml2rfcv3-conforming 
>> xml.
> [JM] And this would be good to capture on authors.ietf.org. I've added a 
> comment to Issue #42 (https://github.com/ietf/authors.ietf.org/issues/42).

I suggest we start with the diagram in Brian's RFC for which I wrote 
some descriptive text (which may need some further tweaks). The problem 
with that is that the text isn't in the actual published RFC.

Another idea is to wait for the next RFC to be published that includes a 
diagram (maybe there are one or more already in the pipeline). The tests 
linked to above also shows we might need to check out some parts of ARIA.

We may definitely need to reach out to screen reader users and 
accessibility experts. I can try to help with contacting somebody at W3C.

Regards,   Martin.

> Thanks!
> Jean
> 
>>
>> The third one is probably the most important of all.
>>
>> Regards
>>     Brian
>>
>>