Hacker News new | comments | ask | show | jobs | submit login
Next thing for Ubuntu to learn (launchpad.net)
82 points by siddhant on May 5, 2010 | hide | past | web | favorite | 33 comments

This sort of thinking strikes me as shortsighted. Ubuntu seems to be getting flack for not having competency in kernel development. And yet, Ubuntu has got to be responsible for driving 10x as much Linux adoption as ext4 or whatever other technical feature you like. There are a lot of people who can write drivers. To date, there is only one company that has even come close to pushing Linux into the mainstream. The benefits of that are very diffuse, but they probably include, yes, more overall kernel developers! Rather than attacking them for not having the wherewithall to crank out kernel patches, we should be thanking them. They bring something to the table that literally noone else has. Even Ted Tso, who is a serious badass, cannot say that.

Hackers working for Google, Red Hat and Novell aren't setting their day to day priorities around increasing Linux adoption. They're interested in making money for Google, Red Hat and Novell.

With Google, I can maybe see your point. But how anybody can argue that Red Hat & Novell's fortunes are not tied to Linux adoption rates, is beyond me.

I guess he means that if it's not a problem in a use case in which their corporate clients are interested, it is a non-issue for them.

Last time I checked much of the work in Ubuntu was done by the teams (Kernel team, Server team, several translation teams, etc), the majority of members of each team are not employed by Canonical, and this is a problem.

As a matter of fact, I know of someone that worked in one of the Ubuntu teams, when his work became well known, and someone at Red Hat discovered that he was not paid they just hired him.

EDIT: I do not see this as a problem though, I watch closely the Dragonfly BSD project and as far as I know the majority of the developers are not paid to contribute with the project.

I don't think there is as much sniping of talent in the BSD world as in linux. I feel like the Linux distros are more competitive with each other so it seems like a bit of a problem to me.

Competition is a good thing. It's only a problem for Ubuntu.

And it's not even that much of a problem for Ubuntu's end users, since Ubuntu's biggest problem is making meaningful contributions upstream.

This reminded me the story of "unpaid internships" (http://www.jonobacon.org/2010/03/19/two-ubuntu-community-tea...), IMHO, there's nothing wrong, but their HR can do it better?

It's a little snarky, but Ted isn't the first to have observed that Canonical has very high developer turnover rates.

One of my friends, through efforts involved in (OSS subsystem) maintenance, worked at, and loved Canonical for a period. He's a seriously talented dude and widely considered number 2 in his field. But yeah, he ended up at another Linux vendor and has been there for a very long time since.

I don't know anything about the personalities involved, and, for all I know, the letter is quite justified; but "I think that I've done you a big favour by giving you 10 minutes of my time" seems to be one of those remarks to which there is [a canonical reply](http://modernarthur.com/blog/christwhatanasshole.html) (no pun intended).

Ted Tso wrote the majority of the filesystem in question, so he is pretty much the authority on the matter.

He was hired by Google, and his point is that if Ubuntu is going to make a distribution, they need to have the engineers available to solve it, instead of just bugging actual engineers over and over again, which I see is reasonable given they sell support.

To be fair, Ted was also involved in a public spit with the Ubuntu kernel team about a year ago regarding an ext4 bug that he tried to "not-a-bug" based on a strict reading of the standard. He ultimately lost the argument, and had to fix ext4. See:


and especially:


Whether that has anything to do with this particular sniping incident, I obviously have no knowledge. But I think it's an important piece of context.

Edit: after reading the whole bug report, it's also important to point out that the issue at hand here is a real and still-unfixed but in ext4 that has been worked around for Lucid. Ted's appeal that no one from Fedora has complained seems a little shallow given that there's a 7-line shell script at the top of the bug to exercise it.

Really, this doesn't seem warranted here. I know the kernel team doesn't like Canonical. But I don't see what they did wrong here other than pick a different workaround than the one Ted suggested.

I dunno, as someone who does a lot of free support, the quote in context sounds blunt but not asshole-ish to me. (Especially if Canonical has a track record of trying to get free work out of him or other kernel developers.)

"I've already done Ubuntu a huge favor by spending ten minutes to do a quickie investigation. Ubuntu needs to learn that it can't rely on upstream developers to jump through flaming hoops on short notice before a LTS release deadline as a cost-saving mechanism to avoid hiring their own senior kernel engineers."

I knew who he was and still thought it was an asshole comment.

Though I was particularly put off by the bit at the bottom, where he says that even if Ubuntu does hire and train up people to work on upstream kernel or filesystem issues Google's deep pockets mean they will just poach them once they are trained up.

Hopefully at Google they'll actually work on the upstream Linux rather than the Android fork.

He's working on ext4 at Google: http://www.h-online.com/open/news/item/Ted-T-so-moves-to-Goo...

(Probably other things too, but I suspect they hired him more because of their infrastructure requirements than for Android)

And to be fair, he said "Red Hat or Google" will hire them. He was implying (saying, actually) that Canonical don't pay their engineers enough, not that Red Hat or Google pay exceptionally.

Asshole comment or not, if Canonical wants help they should pay for it. Ted's not obligated to continue work for free simply because he's done so much work for free in the past. In fact, I'm pretty sure it's the opposite.

I might fix my girlfriend's computer for free, but I make sure to bill everyone else.

Asking him to work for free, and asking him not to be an asshole are two very different things.

I assume you don't act like an asshole no matter who you are helping. I'd go further and assume that even when you turn down people asking for your help, and suggest they might want to hire someone instead, you're still not an asshole to them.

It's pretty hard to not to make "an asshole comment" when you're trying in one email to convince a company to change their way of doing business.

Anyone who has watched Canonical since Hardy Heron (also an LTS release) knows that the company has deep and systemic problems. I wouldn't be surprised if the tone of Ted's response is also a result of past history with them of the sort that's illustrated by this message.

I might go even further and say that if you're as serious a member of the Linux kernel development community as Ted is, you might be rather concerned about Canonical's practices and how they're likely turning off a lot of would be Linux users.

How on earth is Canonical turning off would-be Linux users? They've finally made a distro that has a reputation (deserved or not) for being usable by the common man. How is that anything but good? You think Linux would greatly benefit from being more of a geek ghetto than it already is?

It's not good when they then release broken distributions or upgrades that break stuff that was previously working.

Going back to Hardy Heron, if you were using a machine with a Broadcom 43xx wireless card, you would have been rather upset for that to have stopped working. Or for HH to frequently crash without saying anything (even Windows gives you a BSOD).

What I'm saying is that while the objective is obviously good, the execution appears to be so bad than in combination with that objective it may well be a disaster.

What are these deep and systemic problems everyone keeps talking about? I've seen them mentioned in this thread a couple of times now, yet no one actually mentions what those problems are. I'm curious.

http://news.ycombinator.com/item?id=1256826 is a start for evidence of them.

If what Ted is saying is true about them not paying anyone (who's good) enough to retain them nor hiring a few serious kernel hackers, then that pretty much makes the case, wouldn't you say?

I.e. given that they take a snapshot of Debian unstable and then furiously try to make it work in 6 months.

Another bad and telling sign is that ship dates are more important than quality, and that will also lose them good people. I would not want Canonical on my resume for very long after 6 months before the release of Hardy Heron and I have found such work places to be utterly depressing and demotivating.

T'so is one of the leading developers of the extended file system used by Linux.

Ubuntu surprises me often with their amount of patching upstream projects. Somehow, they can't wait for projects to publish stable releases.

A little digging in the package repositories showed, that 2.5% of lynx packages contain svn, git or cvs in their version number. This means, they were either a stable release that was patched with something from the repo or even a complete snapshot from the repo. Debian, for comparison, has only half as many packages with that criteria.

What Ted seems to ignore is that Ubuntu's company is a barely profitable one so they just can't afford (yet) to have lots of well paid kernel developers.

Canonical does not make billions unlike Google and even Red Hat. It was funded by a guy who gave money from his own pocket and a lot of it and managed to make Linux popular, so kudos to him and I'm sure they'll get lots of kernel developers if they ever manage to be very profitable.

I don't think Ted's ignoring that; it's just not his responsibility. If Ubuntu wants wizard-level fixes at short deadlines, they need to hire and retain wizards, and that's that.

Exactly. There are certain required costs to play certain games, and if they're not willing or able to pony up the resources necessary to play their game (making a good release out of Debian unstable every 6 months) then that's not his problem. I myself have left more than one company because they hired me to do X but then didn't have the will or the means to buy necessary stuff (e.g. a UNIX license).

See elsewhere in this topic where I addressed how their low quality releases starting with Hardy Heron may in fact be hurting Linux. If true, that and the resulting backlash against Canonical would have it end up in even worse shape financially and therefore less able to hire hackers of the level required. I.e. kernel hackers like Ted who keep fielding their requests for help and who are acutely aware of their lack of quality in their releases might be particularly unmoved by the "we're bringing Linux to the masses" argument.

> then that's not his problem

Who said it was his problem?

I'm using it as a figure of speech, not a quote or paraphrase of anything anyone said.

But that is part of the problem, Ted is complaining that he's not responsible for fixing Ubuntu's bugs and then people say he's right. Only problem is that Canonical never said he was responsible for anything. So what is he ranting about?

> If Ubuntu wants wizard-level fixes at short deadlines, they need to hire and retain wizards, and that's that.

Who said they want those? They just don't have the money to pay for such devs, they know it, and that's that. Also who said it was his responsibility? Sounds like a big straw man to me.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact