Re: [Nmh-workers] base64 and text parts

[Lyndon Nerenberg]

On 2013-10-20, at 6:08 PM, Ken Hornstein <address@hidden> wrote:

> Specifically, nmh uses Unix line conventions everywhere.  There are two places
> where it knows how to convert to canonical format: when sending via SMTP,
> and retrieving via POP.  This hasn't been an issue in the past, because
> a) we've never needed the canonical message format ever, and b) at worst
> text will be encoded using q-p and all of the right stuff will happen.

The place where CRLF really matters is for crypto verification, as noted 
earlier in the thread.  However all the UNIX-based crypto tools I have used 
(e.g. gnupg, openssl) deal with LF-terminiated UNIX text input just fine, so I 
don't think we need to worry about how we store decoded text in the filesystem.

The (somewhat artificial) case I would specifically examine is decoding some 
base64 encoded PGP signed text and piping that off to pgp for verification. If 
there are issues with CRLF->LF conversion during decode, that should show it up.

--lyndon