Re: [Bloat] bufferbloat.net is sorely missed.

["Toke Høiland-Jørgensen" <toke@toke.dk>]

Dave Taht <dave.taht@gmail.com> writes:

>> Okay, updated the conversion to preserve the directory structure.
>> Current version is here: https://kau.toke.dk/bufferbloat-archive. The
>
> I get a 404 on that right now across the board...
>
> https://kau.toke.dk/bufferbloat-archive/projects/codel/wiki/Benchmarking_Codel_and_FQ_Codel/

Was moving things around and hadn't set up a redirect yet. Should work
now, and the server also answers to www.bufferbloat.net. I did an
nsupdate, but the linode nameservers don't seem to have picked it up
yet. Not sure how to poke them to do so.

>> overview pages and index are somewhat lacking at the moment, but the
>> wiki pages are there at the same relative URLs as on the old site.
>
> Having the original data available to those willing to work on fixing
> it would be useful, I was reluctant to share the sql dump widely due
> to passwords being in it. Can more of the sql dump land somewhere? Do
> you have other conversion scripts lying around?

This is all in the github repo in the export/ dir - both scripts and csv
files. Did another export from the database, merging several tables with
the information needed for the wiki. There's also an export of the
attachments which I haven't done anything with yet.

> (I used pandoc to convert from textile to markdown originally plus a
> few sed scripts)

Yes, mine are python scripts that also run pandoc.

>> I can update the DNS to point at that server and have old links start
>> working again; if no one objects, I'll do that later today.
>
> Well, to save on cost and improve bandwidth I'd rather it moved to the
> web server lists.bufferbloat.net is on. (and ultimately a cdn). I was
> reluctant to move it there because the redmine integration was
> failing, but for a static web site, no problem.

Have no objections to moving it once we hit something a bit more stable;
but for now, having it on my own server allows me to iterate quicker.
It's at my uni on a gbit link, so should be fine (famous last words).

Having several mirrors and sticking all the addresses in DNS would also
be a way to do things, if we can make sure they all update from git
regularly. Figure these things can be sorted out once we get the
conversion itself done and have it go into maintenance mode (where we
actually take pull requests for the *content*). :)

-Toke