[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Development donations

From: Marcus Brinkmann
Subject: Re: Development donations
Date: Wed, 2 Apr 2003 18:39:31 +0200
User-agent: Mutt/1.5.3i

On Wed, Apr 02, 2003 at 03:36:21PM +0100, Ciaran O'Riordan wrote:
> Could you explain the system a little more?
> For example: I'd like to fund the development of a working
> ppp system for GNU.  Could you make an estimate as to how
> many points this would require?

I don't know exactly how many points equal one hour of work, and I don't
really know how many hours getting PPP to work on GNU/Hurd would require, so
it's really hard to tell.  (You probably notice now that we don't have
much experience with the maintenance point project yet).

For PPP specifically, we had to implement a tunnel device in pfinet (which I
did), port the BSD ppp to the Hurd (which Neal Walfield and someone else
did), and now we need to get the com device in Mach to work at higher
speeds.  I once looked into fixing the com driver, but that requires
separating the bottom handler from the interrupt handler, and I decided that
it was not worth it given that we have oskit-mach which has a cleaner
approach to the problem.  However, oskit-mach required and still requires
a good deal of unrelated, substantial work to get working.

Daniel Wagner is currently writing a serial interface to it.  This is not
only required for PPP, but also for it.  It would probably be faster towards
getting PPP to work to just fix Mach, but that would be "dead code" - it
would do what you want, but it would not really bring us further to a
useful system (it would be just a hack).

I wrote this down at that level of detail to show that a seemingly small and
encapsulated feature can suddenly touch a lot of code, esp in a system that
is as experimental as the Hurd still is.

This is the reason why I definitely can't make binding estimate for something
like this.  Basically any substantial project (like the port to L4, or getting
oskit-mach to work well) is possibly a bottomless sink which can soak up any
effort from the one hour contribution to endless months of whole teams of
developers :), and estimates can vary by orders of magnitutes depending on
how thorough you work on it.

> (If it's a $2000 problem and I have $200 then I would
>  rather pool my money with someone else to get another
>  problem solved.)

Maybe the Maintenance point concept is more suited for lots of small
contributions to add up and be directed towards whatever maintenance work is
required at that time.

I see what you mean, but I am not sure what to suggest in this situation. 
Maybe for a contribution that exceeds the maintenance points system but is
not yet high enough for a real business contract about a project, it's
better to donate some hardware to a needy developer, or make a contribution
to the FSF or FSF Europe.  It's difficult.


`Rhubarb is no Egyptian god.' GNU
Marcus Brinkmann              The Hurd

reply via email to

[Prev in Thread] Current Thread [Next in Thread]