Launchpad Open-Sourced. Now What?
Launchpad, a Web application developed by Canonical for managing software development, was finally open-sourced last week. But with a number of its other products remaining proprietary, what are Canonical's real intentions towards living by the free-software ideology that drives projects like Ubuntu?
Canonical faced criticism early-on for releasing the Launchpad platform under a closed-source license. The company's founder, Mark Shuttleworth, responded by promising that the software would transition to the GPL license when its revenue model matured, but that open-sourcing it at an earlier date would distract developers and risk useless forking of the code.
FUD or pragmatism?
I'm a bit ambivalent about Canonical's endorsement of proprietary development models for its products. On one hand, it seems more than a little hypocritical that the company behind Ubuntu, an operating system built around deep-rooted principles of software freedom, should keep its code secret.
And the excuse that Canonical's ventures remain closed because developers have to be paid is a weak one: first, there are thousands of well compensated programmers at other organizations who work on open-source projects; and second, developers don't always need to be paid to produce quality software. Shuttleworth's insinuation that generating revenue is incompatible with open-source sounds disappointingly similar to the old FUD plied by enemies of free software for decades.
At the same time, launching a project under a proprietary model and transitioning to the GPL later has its merits. In particular, as Shuttleworth pointed out, it helps avoid the insanity of endless forking and bickering among developers that bogs down many open-source projects. In this respect, Canonical's decision to keep code closed reflects a valuable bit of sanity in a software ecosystem dominated by ideologues in need of a dose of pragmatism if they hope to reach the masses.
What's next?
Shuttleworth, a former Debian developer, has spent millions of his own dollars funding Ubuntu and by all indications is genuinely committed to free software. If he decides to keep some code proprietary, he's doing what he truly believes to be in the long-term interests of Ubuntu. But only time will tell whether Canonical's policies will pay off.
It also emains to be seen whether Canonical's other products, like UbuntuOne and Landscape, will follow Launchpad's trajectory from proprietary development to the open-source world. If they do, the company may be making an important contribution to the free-software community by pioneering a new development model where projects wait to mature before releasing their code to the wild.
In any case, Ubuntu users can take comfort in the fact that, with Launchpad now open-sourced, someone can finally write a patch so the application no longer takes an eternity to serve pages.
gt; pioneering a new development model where projects wait
gt; to mature before releasing their code to the wild.
This takes the opposite view of many projects, which uses the concepts of, “Release early, release often” and “Many eyes make all bugs shallow”. By waiting to release the app after it is nearly-fully written, you never give the chance for the community to review the code or structure of the app, pretty much discouraging people from working on the project.
Plus, no one could do that with a large application unless they had the funding (aka large companies or ones with a lot of cash), and without the software being FLOSS, you’d get less beta testers because some would be hesitant to run the software.
“And the excuse that Canonical’s ventures remain closed because developers have to be paid is a weak one: first, there are thousands of well compensated programmers at other organizations who work on open-source projects; and second, developers don’t always need to be paid to produce quality software. ”
Well compensated programmers who are in companies making money with other products. Canonical is on a tight budget and doesn’t have the luxury that Novell or Red Hat has with the positive revenue stream covering base costs.
I understand how by keeping it closed until it’s gain some level of adoption will help to keep from forking it too quickly and splitting the developer pool.
I’m not sure if I 100% agree to the closed-source start then open sourcing it. A lot of it comes down to trust (that they will open source it) which Shuttleworth and Canonical are so far still keeping.
Where does one get the source code for Launchpad now that it is open sourced?
Now
a) people want the build systems and etc. of launchoad open-sources
b) hopefully they’ll be more quiet now about their unjustified and really useless complaints
c) no need to purchase a commercial subscription from launchpad anymore, I probably won’t be renewing mine next year (thank you OSS advocates! you really know what you’re doing.)
d) projects will start using launchpad ala trac and creating their own instances – when one point of launchpad is 1 login, and multi-project collaboration. I can guarantee you these will start popping up, and I guarantee I won’t bother dealing with such projects.
Google has been doing “develop to maturity then release” for years. Only difference might be the sometimes secret nature of Google projects.
To the FOSS zealots, I ask this: Why must EVERYTHING be open source? What do I gain from Launchpad being open sourced? As an end-user and non-ideologue, I just don’t care. My operating system should be open source because it contains all of my personal data and is the software capstone to all of my purchased hardware. My documents and spreadsheet should be in open formats because, again, that’s where my stuff is and I should not be denied access to it.
But Launchpad? Those people who complain about Canonical’s hypocrisy are pushing a political agenda, not a practical one. They are seeking a black-and-white, us-against-them world. It isn’t immoral or anti-consumer to keep one’s source code closed. This isn’t a religion. Reading this incessant pointless bickering makes me want to vomit.
Ultimately, I’m glad Launchpad was open sourced, but only because now that vociferous minority will hopefully shut up about it. Now that they have some free time, maybe they will actually do something worthwhile for Ubuntu rather than expend their energy trying to get everyone else to admit our moral inferiority to their cult and acquiesce to their perfect utopian world of free software.
I’m a Linux user and FOSS advocate.
But I must say that I’m perfectly fine with sacrificing a little freedom if this allows FOSS to prosper financially and hence grow.
Let’s be honest. The ability to see/use Launchpad’s code is hardly as important as, say, the same with respect to the Linux kernel. Now, the latter *is* important.
[…] Launchpad Open-Sourced. Now What? Shuttleworth, a former Debian developer, has spent millions of his own dollars funding Ubuntu and by all indications is genuinely committed to free software. If he decides to keep some code proprietary, he’s doing what he truly believes to be in the long-term interests of Ubuntu. But only time will tell whether Canonical’s policies will pay off. […]
Chris:
Landscape and UbuntuOne are very different than Launchpad. Launchpad is a non-optional service that Ubuntu contributors were required to interact with if they wanted to get involved in significantly contributing to Ubuntu releases. It is significantly important that technical contributors have access the code of Launchpad so they can take ownership of the workflow they are using. Now contributors can start to address workflow inefficiencies in Launchpad that matter to them but were never going to be a high priority to Canonical’s staffed development hours compared to other things.
Looking further ahead, if Shuttleworth is serious about Debian and Ubuntu syncing release cycles more closely…having an open release management toolbase should help immensely because as synced release policy falls into place, that shared policy can be more easily toolized across the tool space. Debian more than any other group has the motivation to take launchpad and extend it with new ideas..ideas like distributed release management.
On the other hand, Landscape and UbuntuOne are optional end-user services. No different than flickr or twitter or dropbox really. These codebases do not underpin an external contributor process associated with Ubuntu. The only people who really have standing with regard to arguing for the licensing of Landscape or UbuntuOne to be opened are paying Canonical customers. How many people pay for landscape or UbuntuOne services? Is there a competing service provider who does have an open codebase that offers Landscape like services with paying customers? How many of those people would pay for Landscape instead if it were opened?
-jef
dragonsnack: source for Launchpad is available at https://code.launchpad.net/launchpad.
Kragil: great point about Google embracing similar practices with respect to open-source. I’ve been critical of Google in the past as a hypocritical pusher of open-source code that keeps most of its own code closed, but maybe it simply is following a strategy similar to Canonical’s, with good intentions in mind. On the other hand, Google still has plenty of very mature applications that remain closed (e.g. Picasa, Google Earth), with no public plans to open them…
Jef: I like your point about Launchpad being a different sort of project with a different target userbase than Canonical’s other software. I agree that there’s more of a push for making Launchpad open-source, since most people who want to participate in Ubuntu and related projects have to use Launchpad. I’d still argue that Canonical will look like a hypocrite and suffer ill-will from some segments of the Linux community if it keeps UbuntuOne and Landscape closed, but you’re right that that present more of a philosophical rather than practical problem, since FOSS diehards could easily choose not to use Canonical’s closed-source applications.
Chris:
The real questions are… how many people are choosing not to buy subscriptions to Canonical’s end-user services because they are closed codebases? And how many current paying customers would contribute if the codebase was available? Those are the only questions that really matters.
I think the history of Red Hat’s RHN and the motivation to open up of the codebase as Project Spacewalk is instructive here. The reality is, paying customers for these services are the only people Canonical is going to listen to…and rightly so. The real problem Canonical has is showing that these services are good value for money and getting enough people to pay for them at all. Once Canonical has reliable.. paying.. customer base.. then the codebases will be opened.. just like we’ve seen Red Hat open Spacewalk open.
Can Canonical show that Landscape is good value for money? They seem very reluctant to communicate anything with regard to Landscape’s popularity as a for-pay service. That should say something significant considering how glibbly they’ll go on record with estimated Ubuntu deployments. They are willing to make estimates about Ubuntu deployments but unwilling to go on record with Landscape paying subscriber numbers…numbers they don’t have to estimate. I guess being able to say we have millions of non-revenue generating Ubuntu deployments out in the wild sounds a lot better than saying with have a exactly 13 revenue-generating Landscape customers.
-jef
Sense of entitlement much? Wow.