Weekly Edition
Return to the Development page
| |||||||||||||
Bacon: Red Hat, Canonical and GNOME contributions
Jono Bacon responds to the GNOME census and the criticisms of Canonical which have followed. "What the report doesnt take into account are upstream contributions that are built on the GNOME platform but (a) not part of official GNOME modules, and (b) hosted and developed elsewhere, such as Launchpad. As such, while the report is accurate for showing code and contributions accepted into GNOME, there are also many projects built on GNOME technology that are not taken into account due to non-inclusion in GNOME modules or being developed outside of GNOME infrastructure."
(Log in to post comments)
Bacon: Red Hat, Canonical and GNOME contributions Posted Jul 30, 2010 14:26 UTC (Fri) by mgedmin (subscriber, #34497) [Link] And here's Mark Shuttleworth's response.
Bacon: Red Hat, Canonical and GNOME contributions Posted Jul 30, 2010 15:58 UTC (Fri) by clugstj (subscriber, #4020) [Link]
Even older than "tribalism" is mud slinging. Mark seems to be stooping to that now.
Bacon: Red Hat, Canonical and GNOME contributions Posted Jul 30, 2010 16:06 UTC (Fri) by AlexHudson (subscriber, #41828) [Link]
I thought Jono's response was open, honest and excellent: Mark's, on the other hand, I just do not understand. His definition of tribalism includes Evidence contrary to my views doesnt count", but the whole discussion currently is centred around a piece of research - i.e., evidence. It reads more like an attempt to hand-wave away the criticism as a distant relation of sexist/racist behaviour.
If I were to buy into Mark's argument, it seems like criticising Google (as another current example) would fall into the same category - apparently only a tribalist could complain that their wakelocks are in some Android repo and not in the Linux kernel. And that seems to me a pretty obviously false argument.
It's not like this argument hasn't been raised before, e.g. Greg Kroah-Hartman's keynote a few years ago at LPC:
http://www.kroah.com/log/linux/lpc_2008_keynote.html
I don't think Canonical get the brunt of this "bashing" in general though - I think Google get it far worse, and IMHO deservedly so in many respects - so for Mark to cry foul seems a bit trite.
Being a member of a community means being responsible, and not committing anti-social acts. On the news page today alone we have plenty examples of organisations who would be deserving recipients of Corporate ASBOs (UKism, look it up ;), off the top of my head Oracle for their treatment of PostgresQL testers being a classic example. It's not tribalism to challenge businesses to be the best participants they could possibly be.
Bacon: Red Hat, Canonical and GNOME contributions Posted Jul 30, 2010 16:18 UTC (Fri) by cjb (guest, #40354) [Link]
> I thought Jono's response was open, honest and excellent
Hm, I thought it could have been more honest. The criticism is "Canonical doesn't contribute to upstream GNOME", and it seems to be an entirely fair criticism backed by evidence, and the response looks something like "You're wrong because we do contribute upstream, it's just using this new definition of upstream that is surprisingly equivalent to the old definition of *down*stream". :)
I think a more honest response would have been "You're right, we've mostly given up trying to contribute to GNOME, and it sucks. We decided that it just isn't worth our time to go through the processes, and we should concentrate on simply writing better free software, and let other people work on integrating it if they like it".
So, in short, I think Jono's response was partially dishonest because it tries to have things both ways -- it tries to argue that there's something "upstream" about throwing code that's only used by Ubuntu onto Launchpad and hoping, instead of just explaining why it is that Canonical is not particularly interested in contributing to GNOME upstream on their terms.
Bacon: Red Hat, Canonical and GNOME contributions Posted Jul 30, 2010 17:05 UTC (Fri) by jspaleta (subscriber, #50639) [Link]
There is a theme to Canonical employee responses I've been seeing.
They try very hard to position Canonical led projects as "upstream" projects as a defence when they are called on on not contributing back. It's a bit of a slight of hand really. Building up projects on infrastructure you control and you hold sole copyright ownership of is not collaboration..not really. Yes technically they are upstream projects.. and yes technically they are open to collaboration. But the reality is they are Canonical owned and Canonical controlled. What people need to see is Canonical be willing to step away from some of that safe single entity controlled project work and make a significant effort to contribute and collaborate to projects as a peer among equals.. not as a controlling entity.
We could use Novell as a contrast. They do lots of their own stuff that..mono for example. People do get cranky about mono as it is a sort of walled garden that Novell has put together and sort of duplicative. They get their fair share of criticism for that. But at the same time Novell is in there doing the contribution up and down that stack to existing projects along side Red Hat and other entities. People are irked by some of their in-house work because its duplicative..but they are also contributing to a common baseline of technology in parallel with their in-house differentiation. I don't think anyone could reasonably argue that Novell is not pulling its weight when it comes to contributing to shared technology projects from the kernel up into GNOME.
Canonical isn't doing that give and take balance that Novell is. They are just doing the differentiated work...and not giving back in proportion.
And its not about size. There are other companies...smaller companies..up and down the stack..that beat the pants off Canonical in terms of giving back when just looking at the raw commit counts..without accounting for size. Just look at the companies ahead of Canonical in the GNOME census. Fluendo, Collabora, Litl... all of them ahead of Canonical in the census..and yet much much smaller than Canonical. Commitment to engage outside of the walled gardens you control in a collaborative way does. Other entities seem to understand that... Canonical...not so much.
Canonical suffers from the same problem Google does in this respect. Except Google actually has the manpower to just isolate themselves and fork an entire software stack and call it a fork and doesn't have to feel defensive about it. Is this a systemic problem for any vendor who wants to compete in the consumer device space?
-jef
Bacon: Red Hat, Canonical and GNOME contributions Posted Jul 30, 2010 18:34 UTC (Fri) by AlexHudson (subscriber, #41828) [Link]
I think you've neatly illustrated why attempting to divide available free software into "contribution" and "not contribution" is ultimately futile: Mono may have issues, but being a walled garden is certainly not one of them. You may not appreciate the contribution in general, but it is a contribution. Contribution isn't black or white, it's a spectrum and entirely in the eye of the beholder.
Bacon: Red Hat, Canonical and GNOME contributions Posted Jul 30, 2010 19:07 UTC (Fri) by cmsj (guest, #55014) [Link]
Mentioning Litl as having lots of engineering time for GNOME is singularly ironic since their OS platform is supplied by... Canonical ;)
Bacon: Red Hat, Canonical and GNOME contributions Posted Jul 30, 2010 19:35 UTC (Fri) by jspaleta (subscriber, #50639) [Link]
I am aware of that. I'm more than willing to give Canonical the benefit of the doubt. If we want to talk about the role of the platform provider we can...if someone is willing to publicly talk about what Canonical is actually providing in terms of engineering support as a platform provider..and what the expectation is as customer of those services when it comes to upstreamable contributions.
Or more to the point. The in-house UI work Canonical is doing. Is Litl using that? Or is Litl differentiating its UI on top of Gnome without leveraging any of the Ubuntu specific technologies...like osd-notify...like libappindicator. Looking at the UI Litl is offering, I don't see them leveraging any of the downstream work Canonical is doing. But I could be wrong. If you have information about the Litl software stack that I don't have... feel free to inform me.
-jef
Bacon: Red Hat, Canonical and GNOME contributions Posted Jul 30, 2010 15:11 UTC (Fri) by Janne (guest, #40891) [Link]
It seems to me that main problem with Linux and free software is that 40% of the time various project/distros fight amongst each other, bickering about who did and didn't do what. Another 40% is spent of bickering about licenses and morality of various licenses. Remaining 20% is actually spent on coding.
You notice that things like "ui-design" or "marketing" doesn't really appear there at all.
Bacon: Red Hat, Canonical and GNOME contributions Posted Jul 30, 2010 15:38 UTC (Fri) by mpr22 (subscriber, #60784) [Link] Regarding marketing: From a highly biased sample of one, I'm tempted to suggest that this is because software developers (both proprietary and Free) tend to associate the word "marketing" with the kind of contemptible corporate claptrap that triggers a "Bill Hicks moment". As for UI design: Programmers tend to make poor-to-mediocre UI designers when it comes to applications for "normal" users, and an awful lot of FOSS that has a UI at all has had it cast in stone before it gets anywhere near a "normal" user.
Bacon: Red Hat, Canonical and GNOME contributions Posted Jul 30, 2010 17:49 UTC (Fri) by kripkenstein (subscriber, #43281) [Link]
> It seems to me that main problem with Linux and free software is that 40% of the time various project/distros fight amongst each other, bickering about who did and didn't do what. Another 40% is spent of bickering about licenses and morality of various licenses. Remaining 20% is actually spent on coding.
If you look at blog posts and tech news websites, then sure.
But in reality, no, most FOSS developers develop. They don't waste time bickering. It just makes for interesting reading in the rare cases that they do.
Bacon: Red Hat, Canonical and GNOME contributions Posted Jul 30, 2010 17:09 UTC (Fri) by tajyrink (subscriber, #2750) [Link]
Interesting comparison to MeeGo - if GNOME is a platform to build UX on, or if GNOME should be the UX: http://www.jonobacon.org/2010/07/30/red-hat-canonical-and...
Bacon: Red Hat, Canonical and GNOME contributions Posted Jul 30, 2010 18:21 UTC (Fri) by ernstp (subscriber, #13694) [Link]
I don't think 4500 commits is anything to be ashamed of, seems like a good start. 70k commits for RedHat, wow! How much of those are actual code?
Bacon: Red Hat, Canonical and GNOME contributions Posted Jul 30, 2010 18:36 UTC (Fri) by rahulsundaram (subscriber, #21946) [Link]
Since only code was counted, the answer would be all of it.
Bacon: Red Hat, Canonical and GNOME contributions Posted Jul 30, 2010 18:50 UTC (Fri) by jspaleta (subscriber, #50639) [Link]
actually translations were counted as well.
Bacon: Red Hat, Canonical and GNOME contributions Posted Jul 30, 2010 19:17 UTC (Fri) by joey (subscriber, #328) [Link]
Commits are a pretty poor measure of work. Even lines of code is probably a better measure. :)
With that said, my experience is that it's not hard for a single person to make 15-20k commits over 5 years. All it takes is some persistence and a commit after every hour or so of work and/or every bugfix.
The more interesting part of the report, to me, was the per-module maintenance map.
Bacon: Red Hat, Canonical and GNOME contributions Posted Jul 30, 2010 22:02 UTC (Fri) by maks (subscriber, #32426) [Link]
Very entangled way to try to promote bizarre outside of the canonical corp. They should start questioning why it never god a real stand anywhere else and ditch it for vastly superior alternatives instead of wasting their very limited ressources.
Bacon: Red Hat, Canonical and GNOME contributions Posted Jul 31, 2010 16:59 UTC (Sat) by lkundrak (subscriber, #43452) [Link]
Never understood why do people like to bash canonical for not writing code. Come on, they're in a different business. Rolling Stones never committed a single line of code into GNOME either, does that mean they suck? Nor did our postman.
Bacon: Red Hat, Canonical and GNOME contributions Posted Aug 2, 2010 11:24 UTC (Mon) by sorpigal (subscriber, #36106) [Link]
Seems to me like he's saying "You didn't count our GNOME contributions that didn't get included into or submitted for inclusion in to GNOME!" - which is both fair in that not all work needs to be part of GNOME to be useful and ridiculous in that they didn't count all of the non-GNOME contributes of anyone else (e.g. Red Hat) either.
Canonical makes a very positive contribution to Linux every time they try to smooth over rough edges in the user experience--and here I mean integration and testing. Where I have a problem with them is where they try to radically warp the underlying technologies. They simply don't have the manpower to push their vision and don't have ideas that are compelling enough to be widely adopted. Upstart was one thing, but windicators is just a farce.
Bacon: Red Hat, Canonical and GNOME contributions Posted Aug 2, 2010 14:29 UTC (Mon) by jspaleta (subscriber, #50639) [Link]
I'm actually having trouble finding a list of Canonical built items that have been submitted and then rejected. Prior to libappindicator... what did Canonical submit for inclusion that got rejected? I've seen the argument made multiple times that there's a history of rejecting Canonical submissions. I'm looking for a listing and relevant mailinglist discussion archives
-jef
Bacon: Red Hat, Canonical and GNOME contributions Posted Aug 2, 2010 16:41 UTC (Mon) by forlwn (guest, #63934) [Link]
They should be more proactive.
Instead of seat down and discuss what was done last 12 months, they should spend more time planning their future projects, trying to converge and deciding about cooperative working models.
Bacon: Red Hat, Canonical and GNOME contributions Posted Aug 5, 2010 20:23 UTC (Thu) by Frej (subscriber, #4165) [Link]
Well it would help if their upstream projects like notify-osd was a good idea. There is a reason it's never going to be accepted (hopefully - i have no influence).
* Makes notifications of out everything. Which countless of research have shown as a bad idea. (Dont interrupt the user because we humas are so slow at context switches).
* The worst possible every interpretation of the mouse pointer. It exists as a mental focus point for our mind. Making stuff _disappear_ when hovering it is backwards and weird and is not used anywhere else.
* Made keyboard/user activated events such as volume/brightness into popup notification way out in a corner. This moves your focus away from the center of your screen. This is an actual regression. In gnome such user/keyboard events are shown as an alpha transparent icon in a single color on the center of the screen.
Hopefully they will learn :-) and everything will be turn out for the better. Appindicator was at least an attempt at a real problem, visual inconsistency - but with boatloads of technology.
|
|||||||||||||