In an effort to get some user feedback and to jump start discussions in the French community concerning what has been proposed by HTML5 with regards to accessibility, namely the alt, longdesc and headers attributes, I started a thread over at the French mailing list Accessiweb a few days ago.

Unfortunately, although there is no official guideline preventing technical discussions on that list, the list moderator asked us today to stop getting too technical because apparently, some people have unsubscribed from the list and he attributes this to our being too technical (which I do not think we were so much but anyway). So we will try and see if we can continue this elsewhere but in the meantime, I thought I would share the feedback I received.

This is an informal summary of discussions but for those who wish to consult the record of discussions directly (in French), the public archives can be consulted here.

Basic information on respondants

The respondants who participated in the discussions were:

While 3 users participated in the discussions, one user did not emit an opinion on the longdesc attribute and one did not emit an opinion on the headers attribute.

While 4 developers participated in the discussions, 2 did not emit any opinion on any of the issues but asked questions or offered information in regard to the discussions.

The alt attribute

Concerning the issue of the alt attribute being optional, all those who replied on this subject (3 users and 2 developers) were opposed and wished for it to be explicated because it is essential for blind users and also, it shows intent on the part of the author on what the importance of the image is.

The headers attribute for tables

For the headers attribute for tables, again, all who responded on this issue (2 users and 2 developers) were opposed to what HTML5 is proposing because in the case of complexe tables, it was indeed considered necessary and it was felt that the Working Group's reasoning was not sufficient to scrap it.

The longdesc attribute

On the issue of the longdesc attribute, the discussion was a little more complicated and slightly divided.

As for the users (2 out of 3), they felt that while longdesc may not be the most optimal way of providing long descriptions (essentially, because it slows down navigation), it could be useful in some cases to transmit information about complexe images. One user however said that longdesc was mostly interesting for graphs and similar types of images because long detailed descriptions of photos were more or less useful to him. So, essentially, the users wished, until something really better and well supported comes along, that longdesc stay.

As for the developers (and this is obviously where things got technical and caused the list moderator to intervene):

One felt that while it may not be optimal and that it would be great if we could improve upon it, arguments to the effect that it was misused or that not many users consulted it was not a sufficient reason to throw it out as it is used by certain users.

One asked why we should not be exploring the "object" element to treat images to which someone pointed out that if object were to work, it would already and that the problem with that was that object was not explicit enough and left it up to the user and technology to guess what the function was.

One felt that longdesc was not efficient and caused problems because it obliged the user to still be connected to have access to the information; it would be like separating the captioning from a video (to which someone pointed out that HTML does not necessarily require the longdesc to be in a separate file).

He also felt that it would be far more efficient to more closely associate the description to the image, using an element that would have the description as a child of that element. One user responded to this that while the idea of associating the long description more closely to the image seemed attractive at first, it would mean getting that information every time, which could be a pain for pages visited often, especially if these pages had several images, so he preferred longdesc as this way, he still had a choice to access it or not.

So discussions turned then to the proposed "figure" element and that is where everything stopped because it was apparently getting too technical.

So, to sum up, users who said anything about it felt that longdesc could be useful and should stay. Others were unconvinced but we did not get a chance to go further in the discussions.


This is obviously not a scientific survey (nor a scientific report) but an attempt to get user feedback and to get discussions started in my community about what is being proposed for accessibility and what the impact may be on users. However, I hope it can be of use to this group and to the issues at large.

Catherine Roy