Add support for rendering uninterpreted/"raw" messages

[stuhood]

Fixes #8.

[stuhood]

Hm: it looks like unicode-width does not ignore ANSI control characters, so raw messages containing ANSI characters do not fully overdraw. See unicode-rs/unicode-width#24.

The top commit resolves this issue, albeit in a somewhat costly way.

[stuhood]

The Parser is fairly hefty, and should likely be lifted onto the State for reuse across messages.

[Byron]

Thanks a lot, all this looks great!

Since VTE is needed for raw messages to calculate their size correctly, I am wondering if there could be a feature-toggle for the Raw part of the envelope. Without it, it's essentially all like it was before without another dependency.

Is this feasible?

[stuhood]

Since VTE is needed for raw messages to calculate their size correctly, I am wondering if there could be a feature-toggle for the Raw part of the envelope. Without it, it's essentially all like it was before without another dependency.

Is this feasible?

Technically vte is required for any Message, since we don't validate that the message strings don't contain ANSI escapes... but I think that it is probably fine to just document that Item::message Strings must not contain any.

But yes: I agree that it would be good to avoid introducing the overhead of the vte parse wherever possible: can add a feature, and will probably also see about making the vte parse conditional on the message type.

[Byron]

Technically vte is required for any Message, since we don't validate that the message strings don't contain ANSI escapes... but I think that it is probably fine to just document that Item::message Strings must not contain any.

It's indeed possible to feature-toggle the Raw variant of the Envelope which would allow everything related to it to only become available if the, say, raw-message feature is enabled. This would then pull in the vte crate. Interestingly, this would then make the vte feature toggle available as well which would allow it to be used for handling non-raw messages as well.

Before doing any of this, I hope it's possible to make make tests pass, so that the new feature toggle can be tested as well using that machinery.

[Byron]

Closing due to inactivity.