KartikPrabhu[keithjgrant]: sknebel: wrapping each comment with a <div class="comment"> and setting it to "display: contents" works. However I think only FF supports this for now
KartikPrabhu[keithjgrant]: yes p-name on img will pick up the alt. But that would have accesibility issues as the comment-author name will be announced twice on screen-readers (I think)
ZegnatYes, it will be announced twice. You could theoretically block the author name text from being read with aria-hidden though. But I am not sure you really want to go to that trouble
[keithjgrant]I think Chrome added display: contents also. That might be the way to go, with a reasonable column width hard-coded as fallback for other browsers
ZegnatI once read an article by a screenreader user who said alt="" is very annoying for these things because people around them would refer to images. And even if they couldn’t see it, they wanted to know the image was there.
[keithjgrant]I wish there was a way to say the mf of element X is associated with the h-* defined by element Y. Kind of like a <label>’s for attribute can reference an arbitrary <input> on the page.
Zegnatalt="" is probably right, upon scrolling through some quick resources. As any description of the person’s display picture would offer redundant information to the text right next to it (their name), and for redundancy reasons alt="" is apparently fine. See e.g. https://www.w3.org/WAI/tutorials/images/decision-tree/ and https://webaim.org/techniques
ZegnatBTW, [keithjgrant], I know some people don’t like doing this but you don’t have to use an h-card on the h-entry and can instead just use u-author
ZegnatSo then you only need to be able to nest that one URL somewhere in the h-entry. There is some discussion about the authorship algorithm that might then even allow matching an h-card *within the current document* if it has the same u-url.
ZegnatSo you could have h-card outside of the h-entry, have u-author on h-entry match the u-url on the h-card, and some parsers will pick that up no problem
sknebelI think that was what zegnat suggested: if there is an h-card in the current document that can provide details, use that instead of fetching an additional document