Re: Microsoft declared "attractive nuisance"; expect list delays

"Stephen J. Turnbull" <stephen(a)xemacs.org&gt; writes: Ideally, I'd want message to discover that the backtrace data that was pasted into the message buffer is binary data, and then message would label it as such in a MIME part, perhaps using a Content-Type application/x-elisp-backtrace. This requires that the backtrace buffer was originally labeled as binary/noconv by XEmacs, but this seems to be the correct thing to do anyway. Would this cause problems somehow? I'm not sure about the implementation though, but it seems more important to first find out if there is a possible solution to the problem. Isn't data recoded when it is copied between buffers? Then Gnus could hook into that recoding, and save what coding was used before it is pasted into the message buffer, and use that information during MIME encoding. If data is not recoded when it is copied between buffers, then I don't understand how things could work unless the data itself keep information about what coding system it is in, but IIRC that is not how MULE works. Are you talking about autodetection? Isn't it possible, instead, to add some code that says the backtrace buffers are binary/noconv? Autodetecting should IMHO be avoided if possible, and for backtraces we know (I believe) that the data must be preserved as is.