Smuxi

Currently the smuxi engine transforms certain substrings such as emoji etc into richer objects for the frontend to consume. Older frontends sometimes can't consume these, and ignore these objects, resulting in data loss.

Sometimes the transformation is specified by the protocol (e.g. Twitter?), but sometimes this is a non-standard extension (e.g. in IRC) chosen by Smuxi - or adapted from elsewhere like mIRC.

(1) For the latter case, such arbitrary transformations are not expected by users. This poses a potential security risk, and therefore should not be enabled by default. For example, cryptographic strings are sometimes given in the format xx:yy:zz:etc and the current emoji regex clobbers these. (But a more complex regex probably wouldn't be completely safe either, and might clobber other things.)

(2) Even when enabled, frontends should not simply ignore things they don't understand, but at least emit a placeholder to the UI to tell the user that something unrecognised is missing.

16:46:22 <meebey> infinity0: Twitter is doing emojis without image support I think
16:46:38 <infinity0> sure, that is part of the official protocol specification
16:46:47 <infinity0> they would be like "a string :) should be interpreted by clients as an emoji"
16:46:53 <infinity0> this fact however is NOT part of the IRC protocol
16:47:04 <infinity0> i don't expect this, as an IRC user

16:46:49 <meebey> ok I think I get your point
16:47:02 <meebey> when the protocol does not support metadata to do something nice to the user
16:47:11 <meebey> dont make a feature that transforms regular text into something else
16:47:16 <meebey> I get the point now
16:47:21 <meebey> and I agree I think :)