[Gnu-arch-users] Re: project hosts, also "truenames"

[Tom Lord]

    > From: "Stephen J. Turnbull" <address@hidden>

    > Do you think it's an accident that Savannah looks like
    > SourceForge, 

Yes.

The Savannah source code was originally forked from an ancestor of
today's SourceForge -- in that sense it is clearly not an accident.

The primary initial motivation for Savannah was objectionable
terms-of-use in the SourceForge service.  In that sense, the forking
of the code was thought to be a forced political decision with,
inevitably, accidental technical (and therefore political)
consequences.

The primary design effort of SourceForge took place during the dot-com
boom, in the context of and by employees of <whatever that company is
called this year>.  It was thus inevitably a design contrained by a
particular world-view of how to make a business in this industry -- a
particular world-view that has since been reasonably well discredited
and may be safely regarded, therefore, as having played out in ways
reasonably described as "accidental".

The initial design was also highly influenced by grabbing what pieces
were available off-the-shelf at the time.

Yes, it is an accident of history that Savannah looks like SourceForge
and that SourceForge looks like it does.


    > or that arch is hosted there?  I don't think either is.  For the near
    > future, I don't think there's going to be much alternative to
    > integrated project hosting, and there are probably substantial
    > efficiencies to be got from mega-hosts like Savannah and
    > SourceForge.

You are obviously safe in your prediction of the near future.   Much
has been squandered that could have had a better outcome.


    > In particular, developers like you who are sick of the unreliability
    > of ISPs and colocs and WimpModems[tm] and volunteer archive hosts are
    > going to jump at the chance to have a stable platform.

I'd have been as if not more happier with a far lighter-weight and
more flexible design.


    > And SourceForge, at least, provides a lot of spokes from the central
    > CVS hub: mailing lists, archives, compile farm, home page.  I have to
    > suspect that centralized management of such a collection of resources,
    > and the data interchange among them, is efficient compared to having
    > projects randomly seeking out those resources.

Efficient compared to what?  Compared to the currently available
alternatives?  Sure.  Compared to what could have been built by taking
a slightly more competent and forsightful approach?  No.  Not nearly.


    > Arch by itself is only one of several large technical
    > innovations that need to be made to achieve the decentralized
    > distributed model you suggest.  It's going to be a while.  In
    > the meantime, I think arch should look at how to present its
    > advantages in the integrated context.

Given that I've spent about a month now just trying to get a tar
bundle to an FTP site, I'm not holding my breath in anticipation of
the ability to make even the slightest tweak.

-t