Re: Moving to the new infrastructure

[Stephen J. Turnbull]

Sam Steingold writes:

 > In my experience, "e-mail - operated bug trackers" (e.g., RT) are much
 > worse than simple mailing lists.
 > specifically, if you use more than one e-mail address (as I do), they
 > turn into an unmitigated disaster because they tend to require that you
 > reply using the same address which was used when you first submitted the
 > issue &c.

Roundup does not require that you use the same address to reply.  It
does require that you register all addresses you use in advance of
using them, or you'll get bounced as a possible spammer.  I suppose
that for infrequent users this would be equivalent.  Roundup *can* be
configured to automatically register the user (with the usual mailing
list style opt-in email confirmation), but I think this would be an
opening to spambots.

I don't know if other trackers have the feature of allowing a single
user to register multiple email addresses, though.  It turns out to be
trivial to implement in Roundup, so it's always been available as long
as I've been following Roundup.

 > specifically, I no longer subscribe to mailing lists that do not offer a
 > web interface to the subscription process (e.g., mailman) because I do
 > not use e-mail to talk to robots.
 > I will find it impossible to use an e-mail-only bug tracker, such as RT.

Nobody is suggesting that the tracker be email-only, merely that it be
tested that way so that email-only users can be confident that they
can use it.

BTW, AFAIK RT does have a web interface that can be used to submit
bugs.  Are you thinking of debbug?

 > also, e-mail operated bug trackers will be infested with spam (and any
 > attempt to filter span _on entry_ will hamper legitimate bug reports).

Filtering spam on entry is not negotiable, unfortunately, because the
reporting address must be public.  Taking reports by email is also
non-negotiable, some people will not use a web interface, and M-x
report-emacs-bug is just too convenient.  So those legit reports are
just going to have to lump it.

One possibility is to put a moderation interface in front of that.
This may be desirable from the point of view of eliminating not-a-bugs
and not-our-bugs from the database as well.  It is easy to write a
Mailman handler to send some traffic on to a specific address.  It
would probably require hacking Mailman's web interface to allow the
moderator to send the traffic on.

No, I'm not interested in volunteering the work for Emacs, but if I do
it for my project I'll try to remember to report back.