On Mon, Apr 25, 2016 at 09:46:02PM +0100, Dominic Hargreaves wrote: > The patch looks good. However the example output looks ugly for me > (particuarly the 100-char one, where most of the patch descriptions > are uglified by a wrapping out of sync with my 80 character terminal).
I see. The lowest common denominator would indeed be 80 characters. The problem with wrapping there is that some of the "normal" lines like intsize=4, longsize=8, ptrsize=8, doublesize=8, byteorder=12345678, doublekind=3 currently go over that, and it seems silly to wrap those. > If we are wrapping for technical reasons rather than aesthetic ones, > then the choice of max length should be based on those technical reasons. > From [1] the maximum line length appears to be around the 1000 character > area. So call it 900 to be safe? I don't think we should wrap more > aggressively than is required to fit within that standard; that way, we > minimise the damage caused by wrapping, when displayed by MUAs which > themselves wrap. I suppose you're right. Not sure if it makes sense to intend the wrapped line at all in that case. > I'm not sure what sort of warning is necessary; could you be more > specific here? We aren't wrapping everything (at least the environment dump and the actual user-supplied text aren't modified by the patch), so it's still possible to hit the limit. I was thinking about documenting this possibility and advising users to keep well under the limit. A better fix would be be to wrap everything, but I don't have a good patch for that. Hacking build_complete_message() could work I guess, but it's not used by all the sending backends (at least _send_message_mailsend(); not sure we need to care about that.) I think I need to look at this a bit more. Thanks for your input, much appreciated! -- Niko