Can't delete patch attachement

[Scott Kitterman]

In https://bugs.launchpad.net/bugs/252236 I wanted to delete the attachments before uploading corrected patches. It didn't work.

I clicked https://bugs.launchpad.net/ubuntu/+source/python-dns/+bug/252236/attachments/321322 to get to the edit screen so I could delete the dapper debdiff. Then I clicked Delete Attachment. Instead of deleting, I got an error:

There is 1 error.

text/x-diff;
  charset="utf-8";
  name="dapper2.debdiff"
Constraint not satisfied

I didn't change anything in the content type box.

If I edit the content type box to just have text/x-diff in it, then I can delete the attachements.

Note that these attachments were originally submitted by email.

[Diogo Matsubara (matsubara)]

Hi Scott,

thanks for the report. I think this is a fallout from another bug, which caused the wrong content type to be recorded when the attachment was submitted by email. Can you still reproduce it?

I think what needs to be done is to cleanup attachments with content-types containing newline characters. (Altough I couldn't reproduce using staging bugs with multiline content types attachments.)

[Scott Kitterman (kitterman)]

I'll let you know if this specific case comes up again, but regardless of if there was another primary problem that exposed this as a secondary issue, I think it's still a bug in it's own right that I couldn't delete.

[Björn Tillenius (bjornt)]

On Thu, Sep 04, 2008 at 09:29:22PM -0000, Diogo Matsubara wrote:
> Hi Scott,
>
> thanks for the report. I think this is a fallout from another bug, which
> caused the wrong content type to be recorded when the attachment was
> submitted by email. Can you still reproduce it?
>
> I think what needs to be done is to cleanup attachments with content-
> types containing newline characters. (Altough I couldn't reproduce using
> staging bugs with multiline content types attachments.)

I also can't reproduce this, but there is bad data in the DB (mimetypes
containing new line characters), which could cause problem, so let's use
this bug to clean up the bad data. After that is done, this bug will
definitely be gone.

    status triaged
    milestone 2.1.10

[Björn Tillenius (bjornt)]

The following SQL query should do the trick: https://pastebin.canonical.com/12728/