[Standards-JIG] avatars

Richard Dobson richard at dobson-i.net
Thu Jul 1 17:54:47 UTC 2004


> Maybe by generalizing the <obj/> of
> http://www.jabber.org/wiki/index.php/XHTML%20Inband%20Images:

Just a couple of things as one of the co-creators of this protocol.

> Sipub:
>  <obj xmlns='http://jabber.org/protocol/iobj#sipub' cid='id1'
>       id='publish-0123' />

> In-band with base64:
>  <obj xmlns='http://jabber.org/protocol/iobj#inband' cid='id1'
>       mime-type='image/png'
>       encoding='base64' >0F58G4c02a90F58G4c02a90F58G4c02a90F5</obj>

> In-band as plain:
>  <obj xmlns='http://jabber.org/protocol/iobj#inband' cid='id1'
>       mime-type='text/plain'
>       encoding='plain' >THE DATA</obj>

> Oob:
>  <obj xmlns='http://jabber.org/protocol/iobj#oob' cid='id1'
>       src='http://www.server.org/theimage.jpg' />
> -> server will return mime type.

This tag was designed to work in a similar way to the object tag in XHTML so
since the attribute for mime type in XHTML is type and not mime-type I think
we should stick with type.

Also I dont see the point in altering the protocol in this way for
supporting avatars, it was designed with a specific purpose in mind as is
shown in the wiki, also one of the main design features that you seem to
have eliminated in your suggestion is its use of a file hash, which means
once you have the file you dont need to re-download it, even if it comes
from someone else, also inband should not be done as these sorts of things
as they should be pull rather than push, i.e. the receiving user must have
the option of rejecting it and certainly shouldnt need to be sent it again
if they already have a copy of that file (even if from someone else). Also
the use of Oob as you suggest is redundant since the user would have just
put the HTTP url in the XHTML message and not needed the inband object
specifier in the first place. Also your use of sipub still eliminates the
file hash meaning even if you have the file in question already when someone
else tries to send you the image or whatever you still need to redownload it
again, when there should be absolutely no need to.

Richard





More information about the Standards mailing list