[Standards] LAST CALL: XEP-0393 (Message Styling)

Marvin W xmpp at larma.de
Mon May 25 14:55:25 UTC 2020


On 25.05.20 15:29, goffi wrote:
> yes that's the idea. Not `styling=inline` because we don't use
> namespaced attributes and thus I don't think we can add an to <body>,
> but something after the body like `<style xmlns="urn:xmpp:styling:0"
> activated="true" />` (or whatever I don't really care), would mitigate a
> lot my concerns (and would also allow a client taking screen reader into
> account to remove formatting). I'm still not found of this XEP, but at
> least I could then live with it.

The problem with that is that the sending client might not even
implement XEP-0393. And if someone is typing "*text*" it is also unknown
to the sending client if that is intended to be styling or not.

I think, annotating that styling is used is far less important than
making it more explicit that styling characters have to be preserved in
the output. It should be clarified that "*text*" is translated to
"<b>*text*</b>" when styling is applied and not "<b>text</b>", same for
all other styling characters (including blockquotes and code blocks). If
this is done properly I see far less potential for issues due to false

Regarding the annotating: There are two questions that are distinct from
each other:
a) Did the sending person expect styling to be applied?
b) Did the sending client apply styling?

They are distinct because:
- A non-supporting client (or one that has styling globally disabled)
may be used by a person that wants styling to be applied to a message.
This is something that already happens quiet often these days: People
know that other clients will display "*text*" bold and therefor add
asterisks, even if their own client does not support that.
- A supporting client might apply styling to text where the sending
person did not expect style to be added (false positives) as I already

However, an annotation will always only answer the second question. It
also does not clarify what should happen if the annotation is missing.

Certainly, the annotation does not hurt, but the only usecase I can see
right now is clients explicitly sending that no styling should be
applied. However, if this happens without the user explicitly
instructing to do so (e.g. because the developer of the client had some
opinion on XEP-0393), some receiving clients may start to ignore such
annotation, making it completely useless.


More information about the Standards mailing list