[Teammetrics-discuss] lists.debian.org solutions and problems

Sukhbir Singh sukhbir.in at gmail.com
Wed Jun 29 10:38:32 UTC 2011


Hi,

> Done that in the past - works somehow.

Yup.

>> 3. Use the mbox archives provided by Gmane (poorest approach IMHO).
>
> I did not really understand why you regard this poor.  Just because it
> is that slow?  Finally it is a monthly cron job we need to run so the
> download speed is not that important.  If it helps keeping your code
> clean I do not see a real problem.

It is painfully slow. So when I tried it with the debian-med mailing
list, it was still running for thirty minutes and then it I stopped it
because Gmane docs say that:

    This interface is a slight CPU and bandwidth hog, so if it's
abused, it will be shut down.

So let's suppose that we have ten lists and they are running for the
first time. That makes it around 100, 000 messages approximately. This
is going to cause Gmane to be busy for a long time and I somehow am
not comfortable doing that unless I ask for Gmane's maintainers
permission and tell him that we are going to do this.

> Or did I overlooked something.
> Possibly also here no e-mail addresses which would make it similar to
> NNTP information-wise?

The email address are not obfuscated in this approach :-) The only
concern is the one mentioned above.

> Perhaps you might like to ping on IRC?  (I would like to add that I'm
> IRC-blind at working hours and can do it only in the evening - but I'm
> very rarely there.)

Yes, I intend to :-) I am going to in fact!

> Perhaps poking on IRC again and keep on working on the other stuff.  If
> you have enough work to do until DebConf we should perhaps delay this
> topic.  If we succeed the solution is perfectly simple if not we use one
> of your proposed workarounds - preferably the one with full information.

We will do it before DebConf so that we have lots of things to show
for and the more we show, the more feedback we will get.

So let me analyze the Gmane mbox archive method a little thoroughly +
get in touch with them on IRC. The HTML parsing is our last resort.



More information about the Teammetrics-discuss mailing list