On Thu, Apr 11, 2013 at 1:51 AM, Joe Dreimann <[email protected]
> wrote:

> No objection to that.
>
> Two questions:
> 1) Was it the error report that got the user to think it is a Trac
> problem? Do we need to amend this?
>

I suspect most users don't look very closely at the content of the error
report. The Internal Error page has a link for opening an issue on
trac.edgewall.org, which populates the ticket description with the user's
Trac configuration. The user only has to click two buttons to create a
ticket on trac.edgewall.org. I suspect that in most cases, the user doesn't
carefully consider where the ticket should be reported, but just clicks the
two buttons to create a ticket. However, we can change where that ticket is
created with a small change to the Trac source.

[image: Inline image 1]




> 2) Should we encourage people using bloodhound to raise all issues to us
> (incl likely Trac ones)?
>

There is some relevant discussion about that in [1]. It appears to be
possible to change where the `Create` button direct to. I tried modifying
the `default_tracker` variable [2], and it appears to work as advertized.
In the case that the reporter has an account on
issues.apache.org/bloodhoundand is already logged-in, the ticket would
be easily created in the
Bloodhound issue tracker. If the user is not logged-in to i.a.o/bloodhound,
they land on the login page, however even after logging-in they are not
redirected to the /newticket page with a populated form. That may just be a
separate issue we need to address to make the error reporting process go
more smoothly.

After changing the `default_tracker` variable, there may still be some
cases that the `Create` button causes issues to be reported to trac-hacks
[3].


> I would say yes to the second one because so far we've always kept tickets
> like that open as a reference and raised one upstream. For users that makes
> our site a single point of contact, and we know what it is upstream that
> affects them.
>
> Cheers,
> Joe
>

That sounds good to me as well. The argument for single point of contact
seems like a good one.

[1] http://trac.edgewall.org/ticket/10898
[2]
http://trac.edgewall.org/browser/tags/trac-1.0.1/trac/web/main.py?marks=55-58#L53
[3]
http://trac.edgewall.org/browser/tags/trac-1.0.1/trac/web/main.py?marks=554#L546

Reply via email to