Hi Marcus:

In message <d80c754a-77da-9827-66f5-7e337eb6af1e@priesch.co.at>,
Marcus Priesch writes:
>Marcus Priesch added the comment:
>
>> Interesting that Marcus' email doesn't show up in the web interface.
>> I did receive it. Ralf did you get it as well?
>
>i got a submission failure email:
>
>There was a problem with the message you sent:
>    'ascii' codec can't decode byte 0xef in position 1451: ordinal not 
>in range(128)
>
>there were some non-ascii characters in the body of the mail

Probably smart quotes. A message is committed in 2 phases. The first
phase creates a msgXXXX.tmp file. Then renames it to msgXXXX.
The .tmp file for your message was still around.

Also we are running the 1.6.1 version of roundup for the tracker.  It
is unlikely to get upgraded unless we put together our own server and
handle mail ourselves etc.

>interesting that there was an failure submission, nosy mail, but the 
>message was not stored in the db ... looks like roundup bug ;)

IIRC that's an old bug:

 https://issues.roundup-tracker.org/issue2550740 
 https://issues.roundup-tracker.org/issue2550540

are probably related. They are from before I had taken over roundup
maintenance. I have no access to the original reporters anymore and
the patches don't make any sense to me. Also they are python 2
patches.  The changes to make Roundup handle python 2 and 3 mean the
patches have to reverse engineered. Internationalization, character
sets etc. are not my strong suit.

>i have the original mail and the failure submission still at hand - if 
>you need if for some tests ...

Can you open a new issue addnig the original mail and failure message
as attachments. I'll try running it through the gateway on my test
system to see if I can get it to break.

It may be too late to get it in 2.2.0 but.....

>> [...] Did you use my second patch (Fixed2) or the original(Fixed1)?
>
>i used the Fixed2 !

Good.

I committed it:

  changeset:   6693:9a1f5e496e6c

it passed CI for all pythons except 3.11-dev. However it's failing
ther with an disk i/o error in the sqlite tests. So I claim it's a
platform and not a roundup issue.

I think roundup needs to get off travis-ci.

codecov is being a pain but it looks like I have 100% coverage for
changed lines so there's that.