Johannes Schauer <josch@debian.org> writes: > Hi, > > Quoting David Bremner (2015-09-18 14:03:20) >> I'd recommend making your own new test, rather than modifying existing >> ones to test multiple things. I'd also recommend using json / sexp >> output for your tests, since the ad-hoc text format is kindof >> semi-deprecated. > > can you take care of that? My goal was actually just to report this bug, not to > spend more time to develop a proper patch for it :) Consider the bug reported. At the moment I'm not completely convinced the patch is correct, never mind the tests. I agree it looks obvious enough, but the test results don't make sense to me yet. > > Also, a related problem occurs when the Content-Disposition header contains > UTF8 characters, in which case the header value gets encoded. Apparently > notmuch does not attempt to decode it. Example mime header: > > --===============7161366892858136962== > Content-Type: application/pdf > MIME-Version: 1.0 > Content-Transfer-Encoding: base64 > Content-Disposition: =?utf-8?b?YXR0YWNobWVudDsgZmlsZW5hbWU9ImJlZ3LDvMOfdW5n?= > =?utf-8?b?LnBkZiI=?= yes, that sounds like a distinct bug.