Re: Concern about new binding.

[Matt Armstrong]

Sean Whitton <spwhitton@spwhitton.name> writes:

> Hello Matt,
>
> Thank you for your message -- there are however a few misunderstandings
> which I'll try to clear up.

[...]

Thank you for the clarifications.

[...]

> No, individuals definitely do not require a full local copy of
> everything stored in the bug system for nmbug to work. You do need the
> git repository containing all the tags, but it is fine to only have
> some of the messages (e.g. only recent messages).

It may be useful to step away from discussing mechanism, and discuss the
problem and use case.

The problem statement as I understand it is this: a person wishes to
monitor Emacs development by subscribing to emacs-devel. They also wish
to know of "interesting" discussions taking place in Emacs bug reports.

Today, the practical options are:

 a) personally subscribe to bug-gnu-emacs and either read every email or
    write sophisticated filters tuned to their personal interests (this
    is the "AI" problem).

 b) rely on the people conducting the bug discussions to move the
    discussion to emacs-devel when "appropriate." This is a manual
    process.

I don't yet see a consensus that mechanism (b) has been inadequate as a
whole, but exploring alternatives can't hurt.


> I suggest thinking of the nmbug tagging as independent of debbugs
> state, at least to begin with. I think they're mostly solving
> different problems.

Ok. I see your point and agree.


>> Which brings me to: if the point is to make certain kinds of bugs
>> more discoverable, adding that feture to debbugs is another option.
>> For example, if the bugs tagged "interface change" were interesting,
>> debbugs could send updates for such bugs to an "interface change"
>> mailing list that interested people could subscribe to.
>
> Well, you'd have to have debbugs mail the entire bug log to that
> mailing list at the point at which it gets tagged, which seems a bit
> awkward. The nmbug approach does not involve sending any messages in
> order to communicate a tagging of the thread.

I don't think debbugs necessarily needs to email the entire history of
the bug to emacs-devel. If it simply supported a way to cc'd bug emails
from that point forward, to emacs-devel, that could be sufficient to
hold the desired discussions in a broader context.

A benefit to modifying debbugs is that it is a minimal change to
existing workflows.

A solution based on notmuch has the drawback that people must use
notmuch. I think that may be too much to ask, for too little benefit (in
this specific use case -- I think notmuch itself is great).