[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Nmh-workers] My .mh_profile
From: |
Ken Hornstein |
Subject: |
Re: [Nmh-workers] My .mh_profile |
Date: |
Mon, 12 May 2014 15:29:03 -0400 |
>when you attached your .mh_profile, thanks to an underlying
>"file --mime-type" invocation, it gained a content-type of
>"message/news".
>[...]
Wow, I didn't really think about all of that. So yeah, a number of
bad decisions all added up there. The profile format _does_ look like
a RFC-822 message (not a coincidence), so it's not surprising that file
ends up doing that.
>so i tried to save it, using mhstore, and mhstore helpfully stuck
>your .mhprofile into my current folder as a new message of very
>questionable validity. (i'm not sure when i'd ever want an attached
>message part of type news saved to my current folder, but i'll allow
>it.)
Hm, looks like this comment is wrong (from mhstoresbr.c):
/*
* Check if the content specifies a filename.
* Don't bother with this for type "message"
* (only "message/rfc822" will use store_generic).
*/
The reality is:
case CT_MESSAGE:
switch (ct->c_subtype) {
case MESSAGE_PARTIAL:
return store_partial (ct, info);
case MESSAGE_EXTERNAL:
return store_external (ct, info);
case MESSAGE_RFC822:
default:
return store_generic (ct, info);
}
>then, running show or mhshow on that invalid message results not in an
>error, but rather no output at all (apart from a part separator).
Yeah, because mhl isn't matching on any headers, and there's no body.
>it's a home run!
A home run in sucking, definitely :-/
Seems like we should do something different in store_generic() (or really,
store_content()). We should restrict that automatic handling to only
message/rfc822. Right now the assumption is there are no other message
parts other than rfc822, external, and partial.
--Ken