Weekly Edition
Return to the Distributions page
| |||||||||||||
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica)
Ryan Paul examines
rumors of MeeGo's demise. "The Linux-based MeeGo mobile
operating system faces an uncertain future amid rumors that Intel plans to
back away from the platform. The troubled open source software project has
failed to gain broad industry support and appears to be slowing down in the
face of weak demand and declining engagement from its backers. Intel
denied the rumors today, saying that it is still "fully committed" to MeeGo
and intends to continue developing the platform while searching for new
partners. Intel's "commitment" doesn't mean much in practice, however,
because the company's development efforts to date have done little to
advance the project. Unless Intel can attract a partner that is better
equipped to produce consumer-facing software, MeeGo doesn't have much of a
future as a discrete mobile platform."
(Log in to post comments)
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 9, 2011 20:28 UTC (Fri) by kragilkragil2 (guest, #76172) [Link]
In hindsight it is easy to see that Intel took a lot of bad decisions (RPM etc) and now there are no devices to speak off and most of the work might be lost soon.
If Meego would have taken a more Maemo route instead of Moblin then at least we would have the N9 and Meego might get devs from Linaro. Even with Nokias decision the current situation would be better.
Not developing proper UIs is another major downside. It might work for Genivi and cars but not for consumer devices that are build by some Shenzhen manufacturer.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 9, 2011 20:43 UTC (Fri) by mjg59 (subscriber, #23239) [Link]
You honestly think that the (broadly uninteresting in most cases) difference between RPM and deb was a more significant issue than changing to an entirely different toolkit at Nokia's behest? I don't the story of Meego makes anyone involved look good.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 9, 2011 23:00 UTC (Fri) by kragilkragil2 (guest, #76172) [Link]
If you want to run everywhere you should use something that runs everywhere. If you want to have a working community you should work with existing communities. Maemo had a community and it had a proven track record of running on embedded devices. Moblin was just Intels toy OS. Just like Meego is now.
If they had chosen more Maemo they could have gotten contributions/commitment from Debian, Linaro, Ubuntu, Freedombox etc. Now if Intel calls it a day the light go out.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 10, 2011 4:26 UTC (Sat) by arjan (subscriber, #36785) [Link]
So Nokia's N9 is based on Maemo... by your reasoning, the community is standing ready to take over when Nokia stops supporting/developing it ??
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 10, 2011 6:44 UTC (Sat) by tajyrink (subscriber, #2750) [Link]
Hmm, not taking part of the argument but yes, the existing maemo/Harmattan/MeeGo community is very interested in N9, and MeeGo CE is already running on the equivalent of it N950, and phone calls should be basically working with free software distributions from day one on N9 as soon as there are easy rooting instructions.
A lot of software has already been also ported from maemo to MeeGo Harmattan (and would of course work on other MeeGo devices as well with a change of packaging deb -> rpm) with the spreading of N950's in the wild and even without them. It's looking very good for N9 for both default software users and people willing to spend more effort on an alternative software.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 10, 2011 10:52 UTC (Sat) by nhippi (subscriber, #34640) [Link]
That's some serious revisionism. Maemo was not exactly flooded with patches from the community.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 10, 2011 14:37 UTC (Sat) by josh (subscriber, #17465) [Link]
Not just patches, but entire community-managed firmware images:
http://wiki.maemo.org/Community_SSU http://wiki.maemo.org/Community_SSU/Changelog
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 10, 2011 15:43 UTC (Sat) by ewan (subscriber, #5533) [Link]
People did try. Nokia were only ever hostile to the community - there were lots of warm words, but the substance consisted of broken promises, FUD, closed components, secret development, and hidden bugtrackers.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 12, 2011 8:10 UTC (Mon) by dsommers (subscriber, #55274) [Link]
You mean hidden maemo bug tracker like this one?
Or was it the MeeGo bug tracker you didn't find?
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 12, 2011 9:13 UTC (Mon) by andresfreund (subscriber, #69562) [Link]
You do realize that much of the core development/tracking happened in nokia internal bugtrackers?
Check some of the somewhat older bugs for examples...
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 12, 2011 13:59 UTC (Mon) by ewan (subscriber, #5533) [Link]
Precisely. The visible Maemo 'bugtracker' was part PR puff and part one-way bug reporting portal. There was rarely any feedback on issues from Nokia engineers, all the actual bug tracking happened elsewhere, even for bugs in the open source components.
As I said - Nokia talked a good game, but they never really lived it.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 9, 2011 23:08 UTC (Fri) by kragilkragil2 (guest, #76172) [Link]
And the toolkit talk is just BS. Most of the stuff Intel developed for Moblin will not run on a phone for years to come. Nokias swipeUI was late, but at least it is great. And Plasma Active stuff also looks really promising. In a few months they build something ontop of Qt/QML that is way better than anything the GTK/Mutter guys did in years.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 11, 2011 15:42 UTC (Sun) by wookey (subscriber, #5501) [Link]
Maybe not 'more important', but certainly significant. It's not the package format so much as all the tools around it. The differences do affect people's choices in what they work on. It's a hell of a lot more productive to work in an environment where you are already familiar with the tools/mechanisms/details. Clearly, smart people can change to other tools, but it's an extra hump that definately discourages.
For example, I know about debian infrastructure, packaging and tools, but not about he RPM equivalents. I'd need a good strong reason to go and find all that out. So it's fairly easy for me to work on Ubuntu as well as Debian, but not Meego, which is a pity because I'd like a project like it to succeed. But in practice the RPMness is enough to mean that I just don't bother.
Call it prejudice if you like (to some degree it is), but the point is that it does make a difference to the available developer pool.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 12, 2011 9:01 UTC (Mon) by juliank (subscriber, #45896) [Link]
But it's probably the other way around as well: If it were using Debian packaging formats and tools, there wouldn't be much involvement from the RPM world. You could say that our world is a bit larger than the RPM world, though, so going in the Debian direction should bring in more people.
But MeeGo has multiple problems:
(1) Per-vendor app stores -- This obviously is not very helpful to attract partners compared to one central app store that is already full with applications. Actually, there is not anything real yet.
(2) Too fast -- MeeGo moves too fast to follow, especially in terms of kernel releases. Being on the extremely cutting edge is not particularly helpful if you have to create a usable product.
(3) Too hostile -- The Linux Foundation and people working on MeeGo are extremely hostile to people wanting to use the produced software on different systems. We have seen this with Smeegol and also on the Debian side.
(4) Qt -- Both Moblin and Maemo were using GTK+ and GNOME technologies, and the developers of those systems know those things. Nokia then switched to Qt, and the developer now need to work with the toolkit from the "other side".
(5) No results -- Despite many release cycles, MeeGo has still not been able to produce a single stable release. Yes, the old Netbook thing might still work, but of the newer code, much is not really useful.
(6) Not neutral -- MeeGo seems like some strange company playground, not open, and definitely not neutral. Rules basically do not appear to apply to Intel and Nokia, only to others (Harmattan is called MeeGo, despite being .deb and not MeeGo Based; x86 builds require recent SSE stuff for slight performance gains on Intel Atom, while excluding a large range of [competitor] chips not supporting it)
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 12, 2011 9:47 UTC (Mon) by boudewijn (subscriber, #14185) [Link]
Qt is just about the only decision that was really right. Nokia had to switch because GTK simply wasn't delivering.
Moving to Qt everywhere was a brilliant strategy. No professional software developer can doubt that Qt is a great platform, which excellent portability which delivers everything to create fast, modern and good looking applications.
Nokia might have lost some of the crowd who still think of Qt as "the other side", but, well, that prejudice disqualifies them as professionals. And they gained a whole new group of developers who were already really competent with Qt.
Of course, Nokia had to spoil that by developing the horrible MeegoTouch toolkit. But in the past couple of months, working with QML on MeeGo has been wonderful. Incredibly good for productivity. I've seen QML based apps on MeeGo that can compete with anything the proprietary competition can show, and those apps were developed in an amazing short timeframe.
Bottom line is if MeeGo dies, Linux and open source has lost its last chance to get into the hands of the consumers. And it's not just Nokia and Intel management who did it: everyone who kept making trouble over deb vs rpm, everyone who kept making trouble over GTK vs Qt, everyone who started fighting MeeGo because it had technology of the "other side", everyone who felt it necessary to complain that the N9 was not running MeeGo when it was announced, all of those people are guilty of the failure.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 12, 2011 11:08 UTC (Mon) by spaetz (subscriber, #32870) [Link]
> Qt is just about the only decision that was really right. Nokia had to switch because GTK simply wasn't delivering.
So how many QT Meego phones will they be delivering? SCNR, I'll shut up.
> Moving to Qt everywhere was a brilliant strategy. No professional software developer can doubt that Qt is a great platform, which excellent portability which delivers everything to create fast, modern and good looking applications.
Others say the same thing of GTK without flushing. I honestly don't think the choice of UI toolkit did anything to contribute to or to prevent the success (=market share) of Maemo hitherto. But the switch cost them enough time to make top management impatient enough to ditch 'the experiment'. "OSS tried that, failed, thank you. Hi Steve B., take a seat."
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 12, 2011 11:28 UTC (Mon) by ingwa (subscriber, #71149) [Link]
>> Moving to Qt everywhere was a brilliant strategy. No professional software developer can doubt that Qt is a great platform, which excellent portability which delivers everything to create fast, modern and good looking applications.
> Others say the same thing of GTK without flushing. I honestly don't think the choice of UI toolkit did anything to contribute to or to prevent the success (=market share) of Maemo hitherto. But the switch cost them enough time to make top management impatient enough to ditch 'the experiment'. "OSS tried that, failed, thank you. Hi Steve B., take a seat."
Hmm, GTK:
In fact, I agree with everything boudewijn wrote. And I would go as far as to say that the if the switch to Qt hadn't been so late, things would have turned out much better.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 12, 2011 12:09 UTC (Mon) by spaetz (subscriber, #32870) [Link]
Hi Ingwa, I don't want to turn this in GUI A is better than GUI B edit war, and I hope it was clear in my previous post that I was not saying GTK is better than QT or vice versa: All I am saying is that I think, Maemo would have been more successful if it stuck with one GUI for a long enough time (that might be sticking with GTK or switching early to QT).
The switch inbetween (and ANY such fundamental switch would have) threw the project enough back to have top mgmt lose patience.
I don't think comments on the quality of a GUI are appropriate as a response to my post here.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 12, 2011 11:49 UTC (Mon) by boudewijn (subscriber, #14185) [Link]
Number of phones: at least one... Which is about as many as there were Maemo phones. You might argue, if you insist, but then you're part of the problem, that harmattan is not MeeGo, but to an application developer it doesn't matter. The api's are the same. Packaging as deb instead of rpm is completely trivial. Anyone who has a problem with that is not a professional.
It was also not the switch to Qt on its own that took time. There were already excellent Qt-based applications for Fremantle that behaved indistinguishable from other apps on the platform.
But doing a new toolkit on top of Qt, then doing another and another, that took lots of time and was not a good idea. Especially when the result, MeegoTouch, turned out to be so bad. It's a pity we didn't have QML earlier.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 12, 2011 12:04 UTC (Mon) by juliank (subscriber, #45896) [Link]
> Packaging as deb instead of rpm is completely trivial.
Not if done correctly, they are entirely different (but mostly equivalent) file formats and conventions.
> It was also not the switch to Qt on its own that took time.
> It's a pity we didn't have QML earlier.
But I didn't take a closer look at it; I wrote my last graphical
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 12, 2011 12:14 UTC (Mon) by spaetz (subscriber, #32870) [Link]
I don't want to spam LWN with more replies. If you want to persue this any further, contact me at Sebastian@SSpaeth.de and I'll continue the discussion.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 12, 2011 15:07 UTC (Mon) by bronson (subscriber, #4806) [Link]
> Bottom line is if MeeGo dies, Linux and open source has lost its last chance to get into the hands of the consumers.
Wrong.
> And it's not just Nokia and Intel management who did it: everyone who kept making trouble over deb vs rpm, everyone who kept making trouble over GTK vs Qt ...
Again, wrong. Good decisions explain themselves. If Meego had switched and then released a tight, usable, stable product, you would not see many naysayers.
The implosion of the Meego project can be blamed on one thing and one thing only: a persistent failure to release a compelling product. You try to share the blame with random and well-meaning devs who question being yanked between deb vs rpm or GTK vs Qt?? It boggles the mind.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 12, 2011 19:48 UTC (Mon) by boudewijn (subscriber, #14185) [Link]
You say I'm wrong, I say I am right... And your mind boggles...
But the fact is, it's unlikely that there will be another chance to get real Linux, real free software, in the hands of consumers. We blew it. The whisper is going round; "Nokia failed twice, with GTK and with Qt. Intel failed twice, with Moblin and MeeGo. Open Source leads to failure. Look to Android, look to iOS -- they were closed and they succeeded. You cannot work with open source projects and succeed."
Yes, all those developers who for years have whined that they cannot conceivably package their project using rpm instead of deb, all those developers who for years have been saying that Nokia choosing Qt must have been a stupid management decision without any technical grounds, all those people who are so partisan for their gtk-based roots they cannot see the value in developing applications in Qt, all those people who are so partisan for their qt-based roots they cannot see the value in gstreamer or tracker, those people who went ballistic at seeing libkcalcore in MeeGo -- they all, through their divisive bickering have had their part in the failure. These were technical people involved in MeeGo, not management.
The people who have engaged in that bickering, are not "well-meaning devs". Or maybe they are, but "well-meaning" is not a positive quality in itself. They might have meant well, but if MeeGo fails, they are just as responsible as if they had meant ill.
Many people think fighting to make sure Qt "wins" or GTK "wins" or "RPM" wins or "deb" wins is important, and so they fight. And while fighting, they are responsible for spoiling the chances of getting free software into the hands of the masses. And of course, their minds are boggled at that idea.
And that's the problem. People have to realize that their pet prejudices do have an impact. It's natural to define your in-group by antagonism, but please pick the right antagonist. Don't go "Oh, Maemo was deb-based, so merging with Moblin, which is rpm-based, and taking rpm is a capitulation to the other side!". And mark well -- "the other side", that's a quote from the comment I replied to.
In the final analysis, yes, partisan people defining "the other" as people from the g-camp or the k/q-camp have to admit to share in the blame. To all of them, I say: grow up and take responsibility, learn and do better next time. If there is a next time. There might not be.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 12, 2011 22:55 UTC (Mon) by bronson (subscriber, #4806) [Link]
You missed a rather crucial point I was trying to make: if Meego had produced a compelling release, it doesn't matter if they used dpkg, opkg, ipkg, MacPorts, or whatever. Developers would complain, sure, but developers do that. Users would still have bought handsets, apps would have been written, and the environment would have thrived. For proof of this, just look at Android's early development environment. It was horrid (still is) but devs put up with it because it works reliably and users like it.
Now, let's assume that Meego had picked exactly the perfect combination of libraries and tools and nobody had a reason to complain about anything... If they still only produce a low-quality, inconsistent, bug-ridden user experience, it will fail.
So, no. I do not believe that people on message boards, no matter how antagonistic, had anything to do with Meego's implosion. They are a symptom, not a cause. If Meego dies on the vine, it's Meego's fault. Period.
And claiming that Meego is our last chance to ever see free software on handsets sounds like ungrounded hysteria. WebOS, Android (maybe Cyanogen, MIUI, Kindle, Baidu, or something else), or even straight up Debian ARM could all fit the same niche that Meego was shooting for. HTC and Samsung have both expressed interest in funding mobile OS development. The sooner Meego disappears, the sooner a real alternative will take its place.
Nokia has been so full of failure for the last decade (outside Symbian of course) that I don't see anyone terribly concerned that they blew it with Meego and Qt too.
Maemo was nice Posted Sep 13, 2011 13:54 UTC (Tue) by man_ls (subscriber, #15091) [Link] I am sad that Maemo's last days were so horrid. It was a promising platform that could have thrived if Nokia had got its act together. But in the end it would not have mattered much, the moment Microsoft bought its way into the company and destroyed it: Intel was not going to build useful devices by its own.But Maemo's demise might still have been useful if we (as a community) had learned one or two lessons about how not to manage a community. Apparently we haven't. I still hope beyond hope that the next time that someone suggests that Red Hat change its packaging format to APT (or Debian to RPM, for that matter, or any similar pointless incompatible change) developers start screaming in horror. Even worse than we did for Maemo. And managers should read Linus' guide to kernel management, especially the section about irreversible costly decisions.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 18, 2011 21:51 UTC (Sun) by Tet (subscriber, #5433) [Link] WebOS, Android (maybe Cyanogen, MIUI, Kindle, Baidu, or something else), or even straight up Debian ARM could all fit the same niche that Meego was shooting for.Could. But let's be honest here. In the real world, "won't" is far more likely. It's simply too late. iOS and Android have got the market pretty much sewn up by now. That wasn't true at the time the N900 was released, and even after MeeGo was started, it still stood a chance. But they needed to get a usable product to market quickly, and they spectacularly failed to do that. That window of opportunity has almost certainly passed. Will we see a properly open Linux for mobile devices? Probably. But it won't be in the next few years. We had the opportunity to make it happen. But like boudewijn said, we blew it.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 25, 2011 20:53 UTC (Sun) by oak (subscriber, #2786) [Link]
> You could say that our world is a bit larger than the RPM world, though, so going in the Debian direction should bring in more people.
Current phones are ARM, not x86 based. Debian is regularly[1] tested & used on ARM. I don't think the same could be said of any RPM based distribution, or could it?
[1] Debian's involvement with ARM started already in the last century: http://lists.debian.org/debian-arm/
I think one of MeeGo's problems was also that it didn't have any upstream distribution that could have provided base policies for packaging, a base system, extra packages and deps for 3rd party apps etc, like e.g. Debian did for Maemo.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 9, 2011 21:25 UTC (Fri) by xxiao (subscriber, #9631) [Link]
well said, can't agree more.
they're doing exactly the same with one more "LF" project now, the Yocto, again shifting from ipk towards rpm due to "customer request", it's another dead end, i hope it will not hurt openembedded too much in the end this time.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 10, 2011 9:02 UTC (Sat) by bye3838 (guest, #52657) [Link]
I don't think "Yocto is shifting from ipk towards rpm". Rpm is just one of the various packaging formats that Yocto supports, just like ipk. Last time I heard is that both the openembedded and yocto communities were working together to align and merge their effort and technology for the embedded linux build system.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 10, 2011 14:03 UTC (Sat) by xxiao (subscriber, #9631) [Link]
yes but rpm is the default, which means it will be more tested/maintained. logically they will keep ipkg the default, while adding rpm for certain customers as another supported format, but they just swapped the order, which in my opinion, has serious consequence, time will tell again, similar to moblin/meego.
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 25, 2011 20:42 UTC (Sun) by oak (subscriber, #2786) [Link]
ipkg is way to install packages, it's not a real package manager (like rpm/dpkg) that supports upgrades with conflict, replacement & versioned dependency management.
I'M SHOCKED Posted Sep 9, 2011 21:12 UTC (Fri) by b7j0c (subscriber, #27559) [Link]
NO ONE saw this coming. NO ONE
I'M SHOCKED Posted Sep 9, 2011 21:36 UTC (Fri) by jmalcolm (guest, #8876) [Link]
Except perhaps you it seems.
I'M SHOCKED Posted Sep 9, 2011 23:54 UTC (Fri) by csamuel (✭ supporter ✭, #2624) [Link]
I suspect your sarcasm detector is broken.. ;-)
Intel denies giving up on MeeGo, but that doesn't mean much (Ars technica) Posted Sep 13, 2011 4:19 UTC (Tue) by ferringb (subscriber, #20752) [Link] Kind of interesting taking a look at the commit rate per month for public OBS; this is assuming their commit hooks haven't started eating mail- stats for gmane match the stats on lists.meego however....
|
|||||||||||||