Hi Mark, On Sat, Mar 31, 2012 at 03:13, Mark Walters <markwalters1009@gmail.com> wrote: > Secondly it seems a little odd that the second patch above changes the > header stuff from 'From to "From" and then this patch changes it back > again. This is probably only a matter of a tidy history but I have to > admit I am confused about why the reply from alternate address still > works (but it does seem to) Good question. These three bugs (alternate address, References header, double headers in 23.2) basically stem from two problems: poor interactions between notmuch and message.el, and inconsistency in using strings and symbols for header names. All the message.el functions usually assume symbols for header names, but when I was writing the new reply stuff I found it easier to use strings. In emacs 23.3, message.el is happy to accept strings. The alternate address bug was caused by the fact that we were using strings in one place and symbols in another, and 'From isn't equal to "From". Rather than changing everything to symbols, I just changed the one place we were using symbols to use strings. But, it turns out that emacs 23.2 isn't happy with strings, which caused the double headers bug. So I ended up converting everything to use symbols. Because we're now consistent in using symbols, and never pass strings to message.el, the alternate address bug is still fixed. All that's a long way of saying: the fix for emacs 23.2 should actually fix the alternate reply bug as well. This probably means I could drop the alternate reply bug patch, and rebase the other two onto master, and we'd still have all the bugs fixed. Is that preferable, to reduce the number of code changes in the history? -- Adam