Attach files via email

Why do we need a command? Shouldn't the presence of most attachments automatically mean this?

I suspect filenames in MIME messages are not required, and need not be unique which makes the suggested command problematic too.

According to RFC 2045 section 5 the name= attribute is indeed optional -- indeed, it's not even mentioned as an example parameter. I have no idea how common excluding it is, though.

As software advances, people will be less likely to know the name of an attachment anyway. For example, I can copy an area of a graphic and paste that area into Apple Mail; it's called "pastedGraphic1.tiff", but the only way I know that is by looking at the source code of the message in my Drafts folder.

... So while I don't normally like blacklists, I think a better approach here is to have a blacklist of attachments that are hardly ever useful for bug reports (e.g. signature.asc).

On Tue, Feb 07, 2006 at 01:37:10AM -0000, Matthew Paul Thomas wrote:
> ... So while I don't normally like blacklists, I think a better approach
> here is to have a blacklist of attachments that are hardly ever useful
> for bug reports (e.g. signature.asc).

It was discussed before, that we should automatically add attachments
to bugs, but it was decided it was better to make it more explicit.
I'll bring it up for discussion again, though, before I'll implement
this. I see a value of making it easy adding attachments via email, but
I don't really like having a blacklist.

On 7-Feb-06, at 2:51 AM, Björn Tillenius wrote:

> Public bug report changed:
> https://launchpad.net/malone/bugs/30225
>
> Comment:
> On Tue, Feb 07, 2006 at 01:37:10AM -0000, Matthew Paul Thomas wrote:
>> ... So while I don't normally like blacklists, I think a better
>> approach
>> here is to have a blacklist of attachments that are hardly ever
>> useful
>> for bug reports (e.g. signature.asc).
>
> It was discussed before, that we should automatically add attachments
> to bugs, but it was decided it was better to make it more explicit.
> I'll bring it up for discussion again, though, before I'll implement
> this. I see a value of making it easy adding attachments via email,
> but
> I don't really like having a blacklist.

Making it explicit means another command for users to learn and
remember, and another potential surprise during the initial learning
curve.

Ideally, we should be able to apply some simple heuristics to make
this Just Work.

Cheers,

--
Brad Bollenbach

On Wed, Mar 08, 2006 at 04:39:04PM -0000, Brad Bollenbach wrote:
> Ideally, we should be able to apply some simple heuristics to make
> this Just Work.

Can you propose a set of heuristics that would work? Note that they'd
need to be pretty simple to promote less confusion than an additional
command.
--
Christian Robottom Reis | http://async.com.br/~kiko/ | [+55 16] 3376 0125

Isn't it rather the other way around? The additional command approach
has to be proven easier than black listing since it has been pointed
out earlier that it may be hard for a user to get a unique identifier
for the attachment she wants be included in the bug report without
examining the unprocessed mail source.

Yet another approach I can think of would be to require a certain
keyword / syntax in the one line description of attachments which are
not to be scrubbed by malone. But this will be something to be
learned and remembered by the user again.

Anyway, this feature is really important and should be implemented one
way or another, rather sooner than later.

Bug reporters seem to attach an awful lot of files by e-mail, when you
respond to them asking for logs. I have to tell them explicitly to use
Malone's +addattachment link, and there are still people who forget to
do that, giving us reports like bug #30272.

Adding e-mailed attachments automatically to the bug, minus things like
signatures and trailing vCards, would be wonderful.

Increasing the severity because a lot of people are attaching files via email, causing extra work for the bug triages since the files don't end up in the bug report.

I've just fallen foul of this. Could multipart mail at
least be bounced or warned about to the sender
rather than just junking the attahments?

Attachments seem to work in the Debian tracker,
so perhaps you can adopt its tactics for handling
MIME.

Interacting with Malone is currently a disincentive to reporting bugs, especially compared with how
smoothly the Debian BTS usually works.

Testing this on staging shows that the fix isn't correct. The wrong mime parts are being picked up when sending an attachment via email.

 * Attachment added via the web interface:
    https://bugs.staging.launchpad.net/launchpad/+bug/192784/comments/3

 * Attachment added via the email interface:
    https://bugs.staging.launchpad.net/launchpad/+bug/192784/comments/2
    Notice how the attachment is displayed as the comment text, and instead of the attachment the file looks like some kind of text-encoded binary.

The list above is wrong (because of what seems to be a bug in comment ordering). Instead:

https://bugs.staging.launchpad.net/launchpad/+bug/192784/comments/3

 * Attachment added via the web interface (correctly:
    https://bugs.staging.launchpad.net/launchpad/+bug/192784/comments/4

 * Attachment added via the email interface:
    https://bugs.staging.launchpad.net/launchpad/+bug/192784/comments/2
    Notice how the attachment is displayed as the comment text, and instead of the attachment the file looks like some kind of text-encoded binary.

 * Binary attachment added via the email interface:
    https://bugs.staging.launchpad.net/launchpad/+bug/192784/comments/3
    Notice how instead of a the attachment, the file in the librarian only contains the file's URL.

re-opening, since the bug isn't fixed

Abel committed a fix in RF, rev 5721, and it works :)

I submitted bug 194576 about something unrelated via e-mail with an attachment and noticed that the attachment shows up in the description and as an attachment. Is this by design? If not should I open a new bug report about this?

On Sat, Feb 23, 2008 at 12:45:35AM -0000, Brian Murray wrote:
> I submitted bug 194576 about something unrelated via e-mail with an
> attachment and noticed that the attachment shows up in the description
> and as an attachment. Is this by design? If not should I open a new
> bug report about this?

Yes, please open a new bug report about this. It is a bug.

Regards,

Bjorn