[BUG] multipart ID of show != part

[BUG] multipart ID of show != part

[Matthias Guedemann]

Hi all,

I found some problems with the multipart/mixed behavior of current
master. I have several multipart mails where the html part is not
displayed and a text/plain attachment is wrongly reported as text/html. 

I have no real insight into the inner workings of notmuch-show but have
found this discrepancy between "notmuch show" and "notmuch part":

(master last commit c4e00ab69a14bfe5756056614a5fb7df8580cd33)

show produces this JSON:

^Lbody{
^Lpart{ ID: 1, Content-type: multipart/mixed
^Lpart{ ID: 2, Content-type: text/html
Non-text part: text/html
^Lpart}
^Lattachment{ ID: 3, Content-type: text/plain
Attachment: ATT00001.c (text/plain)
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe
^Lattachment}
^Lpart}
^Lbody}

whereas "notmuch part --part=2" produces this (the content of the
ATT00001.c file):

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe

the man page states that the part number is equal to the ID, but here
part=2 returns the content of ID=3 and therefore the wrong message
part.


regards
Matthias

Re: [BUG] multipart ID of show != part

[Daniel Kahn Gillmor]

[-- Attachment #1: Type: text/plain, Size: 964 bytes --]

On 05/23/2011 01:25 PM, Matthias Guedemann wrote:
> Hi all,
> 
> I found some problems with the multipart/mixed behavior of current
> master. I have several multipart mails where the html part is not
> displayed and a text/plain attachment is wrongly reported as text/html. 
> 
> I have no real insight into the inner workings of notmuch-show but have
> found this discrepancy between "notmuch show" and "notmuch part":

the part numbering and representations are in the middle of a major
reorganization; the part numbers should be normalized across all uses
(with the exception of decrypted messages, whose part numberings will
change based on a successful decryption) once the overhaul is done.

i believe that the "notmuch part" subcommand itself is likely to go away
in favor of a --part=X argument to "notmuch show".

We will ensure that the part numbering is correct, but i agree with you
that current master isn't there yet.

	--dkg


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 1030 bytes --]

Re: [BUG] multipart ID of show != part

[Carl Worth]

[-- Attachment #1: Type: text/plain, Size: 993 bytes --]

On Mon, 23 May 2011 19:25:01 +0200, Matthias Guedemann <matthias.guedemann@ovgu.de> wrote:
> show produces this JSON:
> 
> ^Lbody{
> ^Lpart{ ID: 1, Content-type: multipart/mixed
> ^Lpart{ ID: 2, Content-type: text/html
...
> whereas "notmuch part --part=2" produces this (the content of the
> ATT00001.c file):

It used to be that the "multipart" containers were not printed in
"notmuch show" (so they weren't counted). We recently began emitting
these, and counting them, but we neglected to update the "notmuch part"
code to count them equivalently.

I've just pushed a quick patch to "notmuch part" to fix this problem for
now.

Soon, we'll have a patch series from Jameson committed that replaces the
"notmuch part" command with "notmuch show --part". That will have the
advantage of implementing both pieces of functionality with one piece of
code, so that skew like this will no longer be possible.

Thanks for the report.

-Carl

-- 
carl.d.worth@intel.com

[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]

Re: [BUG] multipart ID of show != part

[Matthias Guedemann]

Hi Carl,

thank you very much for your effort. Works well now with your patch.

regards
Matthias