public-inbox.org/git updates

public-inbox.org/git updates

[Eric Wong]

This mainly affects the folks following the top-level
Atom feed at https://public-inbox.org/git/new.atom
or over NNTP.

There'll be over 5000K injected messages from 2006 I missed from
the initial import :x

I noticed this while adding gmane:NNNN mapping support to the
search engine:
  https://public-inbox.org/git/20160811002819.GA8311@starla/T/#u

There will still be some missing messages because some are spam.
news.gmane.org remains up if you want to check my work
(please do, because I am careless)


Also, the following two .onions are geographically separate
so it should remain up if the main server at
http://ou63pmih66umazou.onion/git/ goes down.  These two are
on better hardware than the main one:

	http://czquwvybam4bgbro.onion/git/
	http://hjrcffqmbrq6wope.onion/git/

Users without tor installed may use one of the MITM proxies at
run by www.tor2web.org, too.

And yes, I break stuff all the time and often run barely-tested
development code on my server(s) :>


Finally, HTTPS termination for public-inbox.org is provided by
yahns, a Ruby/Rack server: git clone git://yhbt.net/yahns
(more barely-tested code running off ruby/trunk)

Re: public-inbox.org/git updates

[Junio C Hamano]

Eric Wong <e@80x24.org> writes:

> There'll be over 5000K injected messages from 2006 I missed from
> the initial import :x
>
> I noticed this while adding gmane:NNNN mapping support to the
> search engine:
>   https://public-inbox.org/git/20160811002819.GA8311@starla/T/#u
>
> There will still be some missing messages because some are spam.
> news.gmane.org remains up if you want to check my work
> (please do, because I am careless)

Thanks for doing this.

I wanted to try out your NTTP service, but it took me a while to dig
in my inbox to find your announcement of news.public-inbox.org that
hosts inbox.comp.version-control.git "newsgroup".  Is it possible to
make this a bit more discoverable, or there is not enough NNTP
audience these days to warrant such an addition?  I first went to
http://public-inbox.org/git as I expected there may be some pointers
to other instances of the service, where you list the "git clone"
URL of the archive.

By the way, it felt quite strange to see messages from 8 years ago
mixed with more recent messages when I gold Gnus to fetch the most
recent 333 messages (and of course that fetches 333 messages with
largest message numbers, not sorted by "Date:").  I am assuming that
this is an artifact of "over 5k injected messages" bundle that was
added out of order.

Re: public-inbox.org/git updates

[Eric Wong]

Junio C Hamano <gitster@pobox.com> wrote:
> Eric Wong <e@80x24.org> writes:
> 
> > There'll be over 5000K injected messages from 2006 I missed from
> > the initial import :x
> >
> > I noticed this while adding gmane:NNNN mapping support to the
> > search engine:
> >   https://public-inbox.org/git/20160811002819.GA8311@starla/T/#u
> >
> > There will still be some missing messages because some are spam.
> > news.gmane.org remains up if you want to check my work
> > (please do, because I am careless)
> 
> Thanks for doing this.

No problem!

> I wanted to try out your NTTP service, but it took me a while to dig
> in my inbox to find your announcement of news.public-inbox.org that
> hosts inbox.comp.version-control.git "newsgroup".  Is it possible to
> make this a bit more discoverable, or there is not enough NNTP
> audience these days to warrant such an addition?  I first went to
> http://public-inbox.org/git as I expected there may be some pointers
> to other instances of the service, where you list the "git clone"
> URL of the archive.

Oops, I keep forgetting to do that :x  Should be easier to do
now since I cleaned up the footer generation a few weeks back.

> By the way, it felt quite strange to see messages from 8 years ago
> mixed with more recent messages when I gold Gnus to fetch the most
> recent 333 messages (and of course that fetches 333 messages with
> largest message numbers, not sorted by "Date:").  I am assuming that
> this is an artifact of "over 5k injected messages" bundle that was
> added out of order.

Yes, definitely an artifact of fixing that screwup.
For future inbox imports (maybe LKML), I may leave gaps in the
sequence along the way, but I'm not sure how big the gaps
should or could be.

I suppose one could assign NNTP article numbers like line
numbers in BASIC...