Smuxi

Vsyache Puz wrote:

.Net have good internationalization features, and can wrap stream over stream (so one can change encoding without changing code much)

That is an incorrect assumption. If the first stream reads the bytes from the IRC socket in UTF-8, then you can't put another stream on top that tries to re-encode from KOI8-R because the original bytes are already gone.

It only works if the "middle-encoding" preserves the original byte values which most byte into character decoder or highlevel datatype implementations like String in C# doesn't do.

So if one wants this feature to work in Smuxi he would need first to refactor the IRC library (SmartIrc4net) to read and expose IRC messages as byte[] instead of a String. Then Engine-IRC could decide in which character set it wants to decode based on the channel for example.

it is not necessary to expose message as byte[]. One just need to pass encoding into library, and library should do conversion inside (if library have no way to set encoding, it should be reworked)

Vsyache Puz wrote:

it is not necessary to expose message as byte[]. One just need to pass encoding into library, and library should do conversion inside (if library have no way to set encoding, it should be reworked)

The IRC library that Smuxi uses (SmartIrc4net) has an encoding setting per connection and thus the message decoding is handled for all messages and thus all channels the same way.

One need to implement channel encoding support into SmartIrc4net.
Application will set the encoding, and library will return a properly decoded string.

As I am also the author of SmartIrc4net I wouldn't implement such feature myself. Decoding messages differently depending on the channel. If someone else implements this I would happily accept the contribution.