[Members] Re: disconnect from board to active developers

[Leon Shiman]

on Mon, 23 Oct 2006 17:31:32 -0700 Eric Anholt wrote:
>
>> Hi Alex,
>> 
>> On Wed, Oct 18, 2006 at 04:12:21PM -0400, Alex Deucher wrote:
>> > I haven't seen the budget numbers so I don't know how realistic this
>> > would be, but would Xorg ever consider directly funding the
>> > development of new drivers or significant infrastructual updates?  For
>> > example xrandr++ or a real FB manager may have happened years ago if
>> > it had been funded.  I don't want to take away from individual
>> > contributors, but most of us only work on X in our spare time so it
>> > often takes a big ouside contribution or a long period to time for
>> > major needed changes to happen.
>> 
>> Personally, I'd be extremely wary of anything like this.  To avoid the
>> perception of just giving money to your mates (even if it's completely
>> above-board, it does set a precedent that could be abused later), it
>> would need to be:
>>   - not a full-time stipend,
>>   - covered by a mound of paperwork, including regular status work,
>>   - subject to regular overview,
>>   - something the community unanimously agrees on.
>> 
>> However, there are some very important projects that just don't get the
>> attention we need; the corporate body of support is quite narrow, as
>> opposed to the extremely broad attention that projects like GNOME get,
>> and we don't have enough hackers to have a kernel-like system either.
>> 
>> So getting some of the talented community hackers working on projects
>> through part-time funding certainly isn't the worst idea ever.  But it
>> would take quite a lot to win me over at this stage.
>> 
>> If you have a good idea, please submit a convincing proposal to the
>> board.  It would, however, require strong support from the board, and
>> basically unanimous support from the active community members, so we
>> don't fall into the pit of favouritism/nepotism/whatever.
>> 
>> I do sympathise with you, though; it's extremely frustrating to see some
>> very promising projects drop away because the authors didn't have time,
>> or because the company has changed priorities, or something similar.
>> Unfortunately the bounty programs that were run in GNOME, Debian and
>> Ubuntu (among others) a while back seem to have basically been an abject
>> failure[0], because being able to pay small amounts of money (generally
>> less than )to see important work done is a great idea.
>
>I also am quite wary of using Xorg money to directly fund development.
>The cost of full-time, non-student developers is so high that I think in
>almost every case our money would be better spent on things that help
>our development community indirectly, such as throwing us all together
>at a conference with food and drink and network hoses.  I'm at least
>sure that better decisions will come out of conferencing and waiting for
>someone with interest to do it than speccing out some work and hiring
>within our community to do it.
>
>> [0]: Because I think it's fundamentally the wrong model.  Most people
>>      aren't drive-by hackers: either they will write on it anyway, or
>>      need money to live on so they can give up their part-time job
>>      that's getting them through uni, or whatever.  Bounties don't
>>      provide enough money, but they do unfortunately get a lot of people
>>      interested who can't follow through.  The failure rate was
>>      extremely high.
>> 
>>      I was funded by LinuxFund to do the modularisation work between
>>      January and July 2004.  It wasn't a massive salary, and I certainly
>>      worked more than the 20 hours a week, but having only that and uni
>>      to take care of -- no other job -- was invaluable, and it certainly
>>      wouldn't have been done that quickly if it wasn't for LF's support.
>> 
>>      Unfortunately LinuxFund have now tanked quite badly.
>
>I think the LinuxFund model was pretty successful for what (in
>retrospect, at least) I think it was trying to do, which was to bring
>students into development so that they get experience necessary to
>become useful later.  I was also a LinuxFund intern or whatever it was
>called, and though it produced some questionable stuff (anyone used the
>SiS 300 DRI driver, or pre-EXA KAA?), that trickle of funds ensured that
>I got experience that ended up with me being a full-time X guy.
>
>One thing that I think is important about what LinuxFund did that was
>different from bounties (and SoC to an extent) though is that it was for
>an extended period of time and not based on completing a specific
>project.  We (jcollord or local X community and myself) got together and
>discussed what I ought to be working on, then I went and hacked on it
>for a while, then we'd get together again and talk about what I ought to
>be doing next or how what I did should be improved.  During that time I
>was getting taken to conferences and talking with other X developers,
>and getting integrated into the community.  With bounties or straight up
>hiring, I don't think that this would happen.
>
>Note: approximate direct cost of the LinuxFund intern thing was $1500 a
>month for a student, plus overhead plus there was a part-time manager.
>Xorg may have money, but we probably don't have cashflow to maintain a
>program like that.
>

My belief is that a program like this would attract financial support to 
sustain itself. Potential support can be tested. There are sufficient funds 
to begin it.

First, the organization has to somehow decide they want it. There are now 
strongly divergent views on this question. We have so far little success in 
resolving such divergent points of view. Reaching consensus would not be 
immediate. But I think it's a skill that we, as an organization, need to 
develop.

Leon

>-- 
>Eric Anholt                             anholt at FreeBSD.org
>eric at anholt.net                         eric.anholt at intel.com
>