Slashdot Mirror
Software Developer Explains Why The Ubuntu Phone Failed (itwire.com)
troublemaker_23 quotes ITWire:
A developer who worked with the Ubuntu Phone project has outlined the reasons for its failure, painting a picture of confusion, poor communication and lack of technical and marketing foresight. Simon Raffeiner stopped working with the project in mid-2016, about 10 months before Canonical owner Mark Shuttleworth announced that development of the phone and the tablet were being stopped.
Raffeiner says, for example, that "despite so many bugs being present, developers were not concentrating on fixing them, but rather on adding support for more devices." But he says he doesn't regret the time he spent on the project -- though now he spends his free time "traveling the world, taking photographs and creating bad card games, bad comics and bad games."
"Please note that this post does not apply to the UBPorts project, which continues to work on the phone operating system, Unity 8 and other components."
Raffeiner says, for example, that "despite so many bugs being present, developers were not concentrating on fixing them, but rather on adding support for more devices." But he says he doesn't regret the time he spent on the project -- though now he spends his free time "traveling the world, taking photographs and creating bad card games, bad comics and bad games."
"Please note that this post does not apply to the UBPorts project, which continues to work on the phone operating system, Unity 8 and other components."
The Ubuntu phone failed because it's a fucking stupid idea. People want smartphones with a large base of popular apps.
Yeah, not so much.
1. Not solid through US carriers. ...so yeah, there were seemingly no advantages and lots of disadvantages to moving.
2. Focus on low cost hardware; no "flagship phone".
3. Primary benefits were ideological; no new features or distinction over incumbents.
4. No integration with a movies/music/tv ecosystem.
5. Practically no existing market to leverage.
6. Dependency on browser over App Store model.
7. No focus on a migration path.
Linux was a successful exception to this type of effort for ONE reason: it was drop-in compatible with its competition. Linux was user-installable on commodity PCs that people already owned -and- it ran Unix, a popular OS that already had applications and a technical user base that was used to making tweaks to enable their apps to run on various Unix flavors. Linux was just another flavor.
-nomsh
To be fair, the difference between closed source and open source in that respect is that the open source guys will tell you, the closed source guys just won't bother fixing it.
You're basically another APK, only with (slightly) better punctuation.
Il n'y a pas de Planet B.
It can always serve as a bad example.
I think he's probably allowed to talk about himself in his own blog.
Il n'y a pas de Planet B.
But are the resources all that scarce? Seems to me we've gone from "on the internet nobody knows you're a dog" to "on the internet even your dog can be a 'developer' ".
Resources get tight when you want someone to work on your for-profit project for free or at below-market rates. That's entirely normal - the previous hype of all things computer-related is over.
"Transparent" is a shit show that trades on every stereotype going. A man in drag is NOT a transsexual.
To the closed-source guys, it's a feature - it makes it easier to sell the next version (which may or may not fix the bug that annoys you, but that uncertainty never stopped people from buying it). And the new version will be "improved" by splitting off some existing handy features into a new "premium" "enterprisey" version, because "professional" doesn't mean what it used to.
"Transparent" is a shit show that trades on every stereotype going. A man in drag is NOT a transsexual.
That's not why it failed. It failed because there was next to no demand.
At some point we should acknowledge that many people have quit the field over crappy projects that have left a permanent bad taste in their mouth, one job after another with sociopaths for managers or leaders, barely tech-literate "managers" who keep on insisting that the time to do something can be negotiated down without consequences in terms of quality, contradictory requirements, the cult of featuritis even when (not if) it damages the product, lack of a sense of being valued, shitty work environment, and less stressful or more fulfilling opportunities elsewhere.
Even geeks and nerds eventually outgrow all the crap and want a real life. ("outgrow" in the sense that putting up with the bullshit for one more day will send them postal). Eventually the instinct for self-preservation kicks in.
So the guy has found something else to do that he enjoys ... we need more stories that show that "yes, there ARE other options, and you may enjoy one of them more".
"Transparent" is a shit show that trades on every stereotype going. A man in drag is NOT a transsexual.
route and ifconfig always were shit.
route had a horrible syntax. "ip route" now has a syntax which is almost proper english and you can always use the same command and swap add for del to remove the very same route. ... ohh this new idea, who needs it? Lets call it eth0:1 instead!
ifconfig and multiple ips on one interface
iproute2 is a nice idea.
systemd on the other hand started with solving a few problems (dependencies of initscripts, cgroups to assign ressources and reliably detect running processes and kill them if they are not stopping) and grew to a horrible monster. Nobody would object a cool initsystem. I remember the old days, when systemd came around with fancy bootcharts and fast booting and everyone saw it as the next cool thing.
But they did not know when to stop. If this is successful, let's add that, it will be successful as well and we're knowing what we're doing, aren't we?
Who cares about a bug free system, when there is only one crappy phone, where it works? Of course, the bugs should not be too extreme, but still adding hardware support has a priority as well. And bugfree systems are rare. Have a look at the mozilla bugtracker. And there are many serious problems, ten year old platform bugs and so on. But firefox mostly works and that's the important part. Abitious projects do not have the ressources to do everything perfect. And there are always more new bugs than fixed ones. If you're in a market as mobile phones or browsers, you need to keep up with features. HTML5 gets new features like every month. So when do you have the time to fix the minor bugs of the feature from a year ago? You start doing so, when people actually care and report they are having big trouble with them. Or when you got a bit of spare time. But now while the users are complaining, that netflix runs slower as it does on chrome.
"despite so many bugs being present, developers were not concentrating on fixing them, but rather on adding support for more devices."
This could be a generic description for Linux in general. It is hard to get people who volunteer their time to do work (or is it really play?) on things they don't want to.
Writing new stuff is fun. People will do that. Fixing bugs is hard work. It requires effort and thought and understanding. You can't persuade people to give up their time to do that, it's not fun.
politicians are like babies' nappies: they should both be changed regularly and for the same reasons
... you couldn't run the programs on it you expected from the platform. It could have easily attracted a market of more professional users if it wouldn't have tried to copy iOS and Android.
There is a market for something like the communicator with modern hardware. Essentially a device which on the outside is a regular phone, and once you fold it up becomes a portable computer, complete with keyboard.
The market for portable devices with an app-store is already full. However for some reason both Canonical and Microsoft are chasing it on both the portable and the desktop side. Both fail doing so, even though Microsoft should have known better since its Windows CE had a far greater market share than Windows Mobile.
Another problem with Ubuntu's phone OS: its UI bought into the militantly-fashionable idea of eliminating all physical buttons & reducing the phone to a touchscreen for literally everything UI-related. From what I recall (circa summer 2013, at least), it didn't just ignore things like volume buttons for the OS's UI... it didn't even have an API for thirdparty APPS to read their state or react to button-state changes. It was insane.
It's the same reason why Android & IOS (and Windows Mobile & PalmOS before them) never became popular alternatives to universal remote controls, even though everyone has a drawer full of old ones begging for some meaningful repurposed use at this point. A well-designed remote allows you to grope blindly, pick it up, and (at the bare minimum) raise & lower the volume, toggle mute, and pause/play/skip by feeling the shape of the buttons alone. An app that forces you to divert your full attention to its UI egregiously violates user expectations, and basically sucks to use. It's also why so many people have bought a Roku or Kindle Fire instead of a (slightly-cheaper) Chromecast... having touchscreen-control as an OPTION might be nice, but having it rammed down your throat as the ONLY way to control a device results in a miserable use experience.
I approached the (then) product manager responsible for the OS and proposed a fork in their route-to-market. My pitch was to create an OS for devices built for industry that were being sold running Microsoft® Windows Embedded Handheld 6.5 & Microsoft® Windows® CE and are still be actually sold today.
At the time companies developing and selling devices were not investing in developing new MS mobile OS devices cuz the OS is long dead and the cost of creating new platforms (they still sell device platforms that are more than 6 years old). Also MS Windows Phone was consumer focused completely cutting out the industrial sector.
So they were stuck in limbo with regards to a viable mobile OS.
So I pitched the idea of approaching these companies with the idea of creating a common OS to get around their legacy issues, cuz at the time they were just dipping their toe into Android, and security back then was an afterthought for most.
Talk about not knowing anything about their broader target market, it seemed to me the guys making the decisions were hoping to copy what Apple & Google did, slap a Unix label onto it and they would had a hit on their hands. Without any real depth of knowledge about the wider mobile market.
The mass-market consumer doesn't buy Unix, its a niche market (still) outside business.
iOS is Unix underneath but Apple (very wisely) keep that hidden and point the consumer at the shinny features and eye candy. They talk about iOS and give consumer friendly names to stuff like 'Metal' etc...
But if you think about it, that's exactly what Ubuntu is really, yer they build some stuff along the way but nothing substantial. Don't get me wrong I use Ubuntu regularly and its my preferred distro. But it takes different skills to build something new.
Why the hate on systemd? I LOVE having binary logs I can't easily parse!
#DeleteChrome
Are you a sysadmin? That'd be a pretty convenient excuse. "Well I would fix it, but ...".
Confucius say, "Find worm in apple - bad. Find half a worm - worse."
I'd run that through babelfish if I knew what language it was supposed to be.
Confucius say, "Find worm in apple - bad. Find half a worm - worse."
"A developer who worked with the Ubuntu Phone project has outlined the reasons for its failure"
Ubuntu Phone failed because Canonical failed to engage with the developers and didn't do a deal with the telecoms to provide a rich user experience. Like apple did with the original Apple Phone Demo.
From the start of the project, they were a bit behind Android. By the time there was a "product" (to be generous), it was far too late for a new player in the market. The android and apple markets were far too well established. They stood less of a chance than Blackberry did.
"Oh my God. This is terrible. This is the end of my Presidency. I'm fucked."; ~ Donald J. Trump
Swahili, I'm pretty sure.
"So long and thanks for all the fish."
I don't know, but I am pretty sure I can take Linus in a fight, dark alley or not. His wife, on the other hand...
"So long and thanks for all the fish."
I love the fact that you have no other purpose in life than to scan Internet forums for occurrences of your initials, and knowing that I can summon you up anytime I want just by inserting them into a post. It's like the master whistling for his dog. HAND.
Il n'y a pas de Planet B.
Exactly.
Ubuntu being Ubuntu, Shuttleworth being Shuttleworth. T'was ever thus.
Starships were meant to fly, Hands up and touch the sky - Nicky Minaj
When you put it that way - being buggy cr@p was just icing on the sh|t cake.
Starships were meant to fly, Hands up and touch the sky - Nicky Minaj
this was the most common (and an easy) command. Did you ever do something more complicated?
No
Usually one big problem I see with these projects is that it's difficult to both build a phone OS and come out with hardware at a manufacturing scale that allows selling the hardware people want at a price people can afford. Sony has some decently nice hardware involved in their Open Devices project. HTC also has released kernel source code. Maybe it would be valuable to bring the new OS first to one of these devices that already has market share and look into building mobile phone hardware later on in life.
Remember how much we love how Microsoft decided to make a common interface across all platforms and resulted in making the Windows interface (particularly the now-usesless start men) worse? Mark Shuttleworth must have thought to himself. "You know, that strategy is absolutely going to work for Microsoft. And while we have neither the desktop market or the smartphone market, let's try it!"
(And, as a guy who does UI work from time-to-time over a decade... Hundreds of apps that use Qt, but how many use QML? I remember when QML was introduced and thought to myself "Peope only use Javascript because it's the only language web browsers understand. Almost every desktop application on every platform is written in a type-checked compiled language for a reason -- catching more errors at compile time is a Good Thing (TM).. Why do I want to introduce Javascript into my perfectly good C/C++ code?")
-- Political fascism requires a Fuhrer.
Why the hate on systemd? I LOVE having binary logs I can't easily parse!
Can't you just convert them to text files then? I'm not familiar with systemd but I would be surprised if you couldn't convert binary log files into plain text log files.
The Ubuntu phone was a me-too thing. No one asked for it, wanted it, or bought it.
And yet, Ubuntu ran their game into the ground supporting Unity, and for what?
Let this be a lesson to all the me-too, flash-in-the-pan bullshit instigators.
https://www.youtube.com/c/BrendaEM
Not when the thing is hung up and needs to be booted off CD or something and then you look at the logs to see what happened last time. The problem appears to be more logging interval (ie. no logs for your fault at all).
However, it's kind of stupid IMHO to have log files you can't read with simple tools. Yet another example of the different perspective from developers who grew up with MS Windows.
It's just one of a laundry list of reasons it annoys people - the real problem is feature creep before getting the core of it right. Give it a few years after it stops creeping and there may not be anything to complain about.
So assuming you can't just convert them (systemd doesn't include such a tool to read/convert log files?) could you not just augment systemd to output the same data to a text file as it encodes in the binary file? It's all open source.
Reading the article it sounds like the openmoko all over again
http://michaelsmith.id.au
If you can employ a team of developers to do a fork, then yes, but it's easier to either still use it and hope it improves or use the earlier alternatives that it has yet to match.
Not really the issue at this point - a binary log of the event that actually exists is good enough, if annoying, but text based logs on the earlier systems which can append to log files are currently getting the job done better than systemd. Logging kind of still sucks despite having been deemed "good enough" and the binary nature of the logs has often been pointed out as one case of the very different mindsets involved. You've got to be very careful to make sure that nothing can cause a race condition to a binary log and make the entire thing unreadable. Lennart is not exactly a careful sort of guy so that bit of unneeded juggling with chainsaws keeps getting pointed out as an example of systemd being not the sort of init system some people want.
Personally I'm biased due to the rushed implementation of it on CentOS7 breaking a few things - I don't think users should have to unplug their mouse if they want their computer to boot among a few other systemd implementation glitches at when that distro was new. With some of them I had utterly no idea what happened due to the logging halting before the problem, and others I had to backtrack to the problem via other means - disabling services one by one until it worked.
and grew to a horrible monster. Nobody would object a cool initsystem.
A million times this.
If systemd were just an init system, I could get on board.
...and it's still working thanks to convergence. She even started to appreciate the desktop experience ;-)
Which was basically RIM's idea to stay relevant with the Blackberry. Provide professional-level services (whatever they are) and sell to businesses. Microsoft couldn't have won that way either.
"When you have eliminated the unacceptable, whatever is left, however improbable, must be the truthiness" - Holmes
Im not suggesting a fork, just a patch of a couple of lines that you could apply to redirect logging to a text file. You dont need a team for that.
The resource that is scare in this case is CASH.
A shortage of cash for shit projects? I wish that were true, but SillyValley proves otherwise. This won't attract any more funding only because it's way shittier than most. At some point, sanity begins to push back.
"Transparent" is a shit show that trades on every stereotype going. A man in drag is NOT a transsexual.
On the positive side they have been fixing things, but they seem to still be creeping in scope faster than they can fix what they have.
Im not suggesting a fork, just a patch of a couple of lines that you could apply to redirect logging to a text file. You dont need a team for that.
I posted a few times but wasn't really clear enough - there already is stuff like that (that took a hell of a lot more than a couple of lines) from third parties who came in to clean up that little bit of Lennart's mess from outside of systemd.
It's a mindset difference - to Lennart it's perfectly OK to expect the user to type in "journalctl -b" on a running system to read a log instead of providing a log file that someone can read after booting from other media if they want to find out why the system wasn't starting up, and also he sees a few things that would normally be logged as not so important (hardware faults are not so important if your objective is developing code). He spent all that time writing a snazzy little log reader as a cool and trendy bit of development work instead of doing something very simple and more useful when things actually go wrong. When questioned about it his response was "We are writing an OS here for the general purpose, not just a toy for a clique of kernel developers." So it's done, he's moved on, he doesn't want to change logging so its either has to be done from the outside (which works to an extent) or a fork adding a lot more functionality in that area.
This really does seem like a fundamental problem with the community or development methodology. If one person can completely upend the system to the point where it requires a team of developers to make something like logging work again that is a very bad thing. I hate to think about the myriad of other things that would end up in the "too hard" basket if somebody or some corporation decided to go in a different direction.
Not exactly "one person" - one person with RedHat completely and firmly behind him - and even then others have chipped in to get syslog to poll Lennart's logging system to produce readable files that can be used even if the system won't start so it's not all bad.
The other distros just don't have the resources so accept whatever work RedHat does.
Well I always see the sort of thing that you wrote, that it's all Lennart's fault, not RedHat. Regardless it means the desktop Linux community is controlled by corporations like RedHat because, as you say, the distros do not have the resources to actually maintain themselves without RedHat.
If it was only him it would be just one of a few choices of sound system and init systems with nobody bitching about having his ideas on what linux should be forced upon them.
We can't blame him especially for his early stuff - RedHat management have plenty of options about who they can have running their projects - it appears he impressed them with his "vision" of not being content with just an init system like the upstart people were doing, but an entire takeover of linux and a change to an MS style environment (as you can read on his blog). Without RedHat he's just a guy with ambition who would have to "play well with others" - with RedHat he can act as he has acted. For example, the "if you want gnome you need to have systemd" deal could never have happened without RedHat behind him.
It's one of the reasons I migrated a lot of stuff to FreeBSD (I haven't seen linux crash so much as it has with recent distros - not even in 1995 on the bleeding edge) including now a couple of desktop systems and a laptop.