Slashdot Mirror
SourceForge Drifting
Zocalo sent us a story running at FSF Europe talking about SourceForge's Drifting. Talks about the fact that they are releasing a closed-source version of the code commercially and various copyright related things. Obviously VA owns both SF and Slashdot so I'm skewed, but my personal opinion is that VA is doing what they need to do to make a buck while still providing the SourceForge.net website to the Open Source community. And I think their decision to sell a closed-source proprietary version of the code would be hypocritical, except that they aren't a 100% open-source company any more. And *that* is the part that makes me the most sad.
It doesn't appear Slashdot has much to do with the open source community lately either. If it really makes you sad, consider what blather makes headline on Slashdot. I don't mean this as a troll. I'm honestly disappointed over the last few months about drift on Slashdot too.
I don't think they're concerned about possible violations, I think that sourceforge and VA are going to do everything in a very straightforward, legal way.
What they're worried about is what sourceforge can talk people into doing. Like signing over their rights to code. There isn't any GPL violation if an author wants to sign his code over to someone else, but all the same it's a loss for free software. They're also worried about the fact that sourceforge seems to have hooked people on a particular development methodology, (and a GOOD one at that) but that now since they're drifting away from free software, they may bring developers with them.
I think that the FSF's concerns are totally valid. I think it's really odd that most of the other posters in this discussion haven't even examined the software angle, and more importantly the freedom angle - they're more interested in speculating on VA's *business model*. Since when did all of the hackers that used to read slashdot morph into business students? (And poor ones at that)
-- Truth goes out the door when rumor comes innuendo. -- Groucho Marx
most succesful endevors in the IT world have had both open and closed source components
For example who? Microsoft would be the most successful, they don't have a lot of open source...
If this is the case, then why does SourceForge have a job listing on their jobs page asking for a DBA that can "oversee and deploy the transition from Postgres to Oracle"? The listing also mentions "[keeping] the databases functional for the 200,000 users that use the site on a day-to-day basis." Why would a new version of SourceForge, which they are apparently selling as standalone software, already have 200,000 users? The job listing is specifically listed as "SourceForge.net Database Administrator", not "Sourceforge 3.0 Enterprise Edition DBA/Developer" or whatever.
i d=7438&group_id=1#dba.
Doesn't this indicate at least in some way that they're planning on ditching PostgreSQL for (gasp) non-free software?
The exact listing can be seen at http://sourceforge.net/docman/display_doc.php?doc
This stuff was brought up on the PostgreSQL lists. I'm hoping Tim Perdue will chime in, as he apparently frequents the lists.
J
That's a very interesting idea. I'm just wondering how well development efforts can work without keeping the project somewhat centralized. Imagine for a moment a world where the Linux kernel doesn't have Linux and thousands of people are all releasing their own little patches to the code in a thousand different places. Seems like total chaos.
I do a lot of development using tools like CVS and even with a small team you can occasionally have problems with people forgetting to check in and causing conflicts. A massive web of people all working on their own tangents would turn that into utter chaos.
Seems like maybe the secret is some sort of very robust diff and dependancy tracking tool. If each patch could keep track of what patches it is dependant on, then when trying to apply a patch, it could inform you of dependancies it has an automatically get those patches. Of course I don't think this scales too well. If each patch has to keep track of patches it is dependent on then over time the patch files turn into huge lists of dependancies followed by a snippet of code.
I like the concept but I don't know if it's feasible...
This sig has been temporarily disconnected or is no longer in service
In-house development in large companies can often run into dozens or even hundreds of projects. Aside from the complexity of individually managing all of those, centrally locating all projects to have one interface, one set of tools, and one locus of control could be seen as very positive in a lot of organizations. Also, whenever large-scale software development is going on, most of the disparate pieces are really quite related. For example, when developing evolution, they developed loads of libraries that did specific things. Sure, they were written to be generic, but their source trees, release schedules, build frameworks and so on were still built to be related to the original project.
For example, there isn't a GNOME CVS. You can't check out the source code for GNOME. You check out the source code for about 50 different projects, all of which put together make up GNOME.
All of this is kind of silly though. Let's let VA worry about how to market their own stuff. IANABA (I am not a business analyst) and my guess is that 99.99999% of the people on this site aren't either. Leave that guessing for the bean counters. Far more interesting in this story are the implications for free software. The FSF has a lot of good points that they raised in their article, most of which are being wholly ignored by the threads here today.
-- Truth goes out the door when rumor comes innuendo. -- Groucho Marx
People I work with have been in contact with VA regarding deployment of SF Enterprise Edition...
The cost comes out to ~ $300K/year for 900 users... Not very economical, expecially considering I have an 80% functional install running off of the last (GPL'd) release - v2.61
my $.02
I see some posts from people who are basically asking, "What's the big deal? They're just doing what they need to survive."
The fuss is that for the last few years ESR et al has been CathedralBazaaring (if you'll pardon my verbization) this idea that Open Source software actually makes MORE economic sense than closed source software, because you get the benefits of the "community". Source Forge has basically rejected this idea, and said "screw this ivory tower theory, it's not working and we need to make money".
I'm not ready to declare the experiments a total failure. I believe Stronghold does pretty well with their commercial version of Apache (not sure though), and IBM is certainly putting a lot of effort toward open source. Of course, IBM is hoping to sell hardware, so it's not quite the same.
In fact, ESR has been pretty quiet lately. Considering he was a board member of VA, has he put out any opinions on this move to closed source? Has he resigned from the board?
Sometimes it's best to just let stupid people be stupid.
29,000 projects are currently being hosted on Sourceforge. Okay, a lot of those are vaporware, but I think it's fair to say that there are at least a thousand interesting and valuable projects there. It would be a huge loss to the open-source community if all of these projects were suddenly homeless.
Sourceforge has done more to increase the sense of community among open-source developers than any other site. Whenever I want to find out if someone is developing source code that does something I want, where do I turn first - Freshmeat? Nope, Sourceforge, because it's so convenient and standardized. I know how to navigate Sourceforge quickly to download the latest release, browse the CVS archives, or check their bug reports - whereas all non-SF projects have these things in very different places, if at all (how many other projects have a working CVSWeb up and running - not many!). Hosting a project on Sourceforge makes it convenient for developers to examine what you're doing and join in, which is what makes open-source work.
I never would have joined if Sourceforge was not free, but if it came down to paying a subscription (in order to host a project there) or letting Sourceforge die, I would pay for it in a second. If they have to do this, it would be nice if they set up a system for micropayments - so grateful users could easily donate a couple of bucks to keep their favorite projects on Sourceforge.
I decided to let go of the wreckage called VA Linux about 2 1/2 months ago, so I'm not involved.
The switch to closed-source and Oracle is a result of the... shall we say... lack of success in the sales department while the open source code was available.
> while the open source code was available.
Is it really *supposed* to be gone!!?
I just downloaded SF 2.5 last week because a client wants a site based on it. I look at the project page now and for the last few days there have been no files there!
Glad I got it when I did!
I would prefer a network of sites that hosted a couple of hundred projects each, and had a crosslinked indexing system.
An interesting alternative might be something based on the music distribution software. But does that require that there are enough people interested in the unpopular and new projects? I don't play in that sandbox, so I don't know. How hard is it to find an infrequently asked for title?
Sunsite was the old standby. ftp is basic, but not the same thing. (And that does require central sites.)
The basic flaw is centralization. Sourceforge by being a central point, became a weakness. Any replacement should try to avoid that flaw.
I think we've pushed this "anyone can grow up to be president" thing too far.
Chris DiBona
Co-Editor, Open Sources
Open Source Program Manager, Google, Inc.
The following post isn't meant to be a flame. I'm just sharing what I knew to be true about my own experiences with VA regarding System 12, the embryonic form of what later became SourceForge. Hell, we even came up with the name "SourceForge" back in May of '99..
To quote from the article:
Finally, VA Linux[1] has become rather underhanded in their attempts to grasp exclusive control of contributors' work.
What sold VA on our idea, originally, was that the company was ultimately going to be in a position to assert a great amount of influence over the design of people's apps in the end. In the months prior to System 12's inception, I was asked by Trae McCombs to provide what amounted to a "proposal" he could hand to the people calling the shots, to justify setting up a box for us and others on the team to work with. The details of that proposal went something like this:
System 12 was going to offer "components"... Nice bits and pieces of graphics, sounds, and code that could be fused into pre-existing Linux apps, and perhaps more importantly, used to build new ones from scratch. The idea was to make the Linux developer community dependant upon System 12. Originally, the primary benefit of this was that all Linux apps would have had a similar behavior and appearance, and i'm sure we'd all agree that such a thing was good--But later, a more interesting benefit emerged, in that we (as System 12) and VA, as our parent, would be able to dictate how people were to develop their apps by controlling the components these apps relied upon. We didn't want to view the project that way -- Asserting control was a secondary benefit. VA viewed it as a primary benefit.
Needless to say, Management at VA apparently liked the idea. They liked it enough to set up a dual P3/500 with 50GB of space on it, sitting on a wide open T1. An enormous machine by 1999 standards.
Essentially, VA would have been able to express their desires as a company via your apps. To this day, VA views SourceForge as a tool to advance the interests of the company. Suppose your code relied upon a component provided by System 12. At any point, VA could alter the structure of that component so as to make your code behave nicely with VA-produced software (ala Internet Explorer & Word), or more amusingly, run a banner ad at the bottom of your apps. This was our idea, and its the idea we sold VA on. System 12, the base predecessor to SourceForge, was designed to exert a measure of control over the direction of Linux application development, SO AS TO BENEFIT THE COMPANY. We wanted to become powerful enough as a central development resource that VA would have some interest in hiring us on as permanent employees versus community volunteers. That never happened. We got shoved off the map before we knew what hit us.
Rather than letting us continue development, they essentially co-opted us, and put pre-existing VA employees on the task of developing the idea. "Grow the garden to attract the bunnies, then lock the gate to the garden and sell rabbit meat." The gate got locked a month or so ago when VA announced they're moving SourceForge into proprietary waters.. Soon, (if not already) VA will trying to co-opt those who participated in the garden. I tried to warn you guys, but nobody listened. I got called insane instead, for suggesting VA had something other than purely altruistic motives. I used to be just as big a flag waver as you when it came to VA, but I learned my lesson fairly early on in the game. I'm afraid the rest of you are just now getting a taste of the same lesson we learned.
To milk the community for the gain of the company was part of the plan from Day 1, folks.
You would be VERY wise to move your project and your work off of SourceForge as soon as possible.
Bowie J. Poag
Also, Bowie's original idea was for a widget repository, and frankly, we never stopped him from doing it. SF and the sf name came from other places than from Trae and Bowie (I regisetered the domain name). Bowie is under the mistaken impression that only Bowie can have an idea. IF you do a search on bowie and I in slashdot you'll see how long he's been asserting things that simply aren't true. Also, It was called system12 . He has a new project which also probably won't produce anything called system26 or whatever.
Mr Poag did not have access enough to t.o to destroy it, we took down the propaganda specific stuff ourselves and dropped his tiles into the resources set.
Chris DiBona
Co-Editor, Open Sources
Open Source Program Manager, Google, Inc.
As such, it's perfectly valid and reasonable to say "this open source stuff isn't allowing me to compete in this capitalistic market!" The context would be something like SourceForge, since for industries less about software authoring, the "open source stuff" can still be a way to cut costs and own outright the means of data processing instead of renting it (say, from Microsoft.)
However, the problem comes when people don't even think to ask any of the other questions: foremost, I think, would be
Is this capitalistic market itself imposing power relationships on me that aren't to my interest?
THAT is the relevant question. Look at the big picture... look at the types of power relationships that exist among vendors, users, developers... it may be that Open Source never does make a sensible business model, but in a world where 'sensible business models' amount to serious power inequities between players and a Darwinian reduction of industries to only the most aggressive, restrictive players, is a business model really the thing to want? If that is the game (and with Microsoft being found repeatedly totally guilty of power abuses and wrist-slapped cautiously, I suggest it is), is it even proper to consider only how best to play that one particular game?
Microsoft knows what it's dealing with when it makes Open Source and the GPL in particular, public enemy number one. These are not effective economic weapons- they are effective specifically at breaking the hold a restrictive vendor exerts on its victims/developers/customers. If you can have ownership of your own software you can't be armtwisted- you are immune from power abuses.
This is in a context of business, again, and power abusers have the most effective business model IF most people are subject to their power. People using open source or developing it may never, ever have comparable economic power or competitive business models- but they can wield a 'spoiler' effect, allowing others to bail out of the proprietary sphere if it's getting too restrictive for them. This is what threatens Microsoft, not some notion that Red Hat will end up with a billion dollar war chest.
And it is right for this sort of thing to frighten power abusers- because it is in fact antithetical to their primary business model. If they were just selling service and quality and working hard it'd be another story- but the winning strategy has been to twist power relationships for all they're worth, and that is precisely what is threatened.
How does all this apply to VA and SourceForge?
Well- they have a choice, though it may be already made for them. They can go the one way- keeping open, and losing in the marketplace but enabling a wide spectrum of 'spoiler' projects that keep proprietary software in check. Otherwise each project will have to maintain its own web presence at its own expense, as I do (Mastering Tools). Or, they can roll the other direction, increasingly twisting power relationships to compete in the marketplace on the marketplace's own terms (even if those are set by hardcore libertarian ideology and best illustrated by Microsoft). If they do that, though, Free software itself is a threat to them, because it destabilises power relationships and makes it possible to avoid lock-in.
It sounds like they're doing the latter. Pity- I guess they felt they had to grow grow grow, to compete in the marketplace and maintain stock valuation. Unfortunately, for them to take this approach is antithetical to free software itself, so I would say they are fucked.
Step 1: Start an open-source based company
:)
Step 3: Profit!
Okay, how about doing this in true open source fashion from them folks. Just an idea, probably won't ever happen.
First off, what Sourceforge has today is a unique combination of services that nobody else I know of has. Sure you can buy web hosting and even FTP serving pretty cheap, but not with full cvs, bug tracking, and built in message boards. Certainly they also have the mind share out there as the place to host an open source project. For all the downsides, this should be a powerful combination.
In my mind, they should do everything they can to keep the tools used on that site open source. At the same time, they should be charging a nominal fee to those folks wishing to host their project there. Heck, it wouldn't have to be much. Figure it like this....
29,275 projects now hosted
$10/month for each project
$5/month additional for mailing list
$1/month for each person authorized to commit code to a project
Let's figure that they only retain 2/3'rds of the accounts in doing this. Of these, let's say about 1/3 add in some features of some sort. The mailing list thing was simply an example.
19,321 projects left
6,376 add in about $7 in features
$193,210 in hosting charges
$44,632 in feature charges
$237,842 total billable each month
$2,854,104 billable annually
This in my mind is a win win for everyone. Sourceforge charges a very reasonable fee for services, and they can show the project off as a profit center, all the while selling their proprietary version on the side. The dead projects that have long since lost developer interest vanish, or are picked up by someone else. Heck, if a project is truly interesting now we'd have a way to get non-developers involved by helping fund their hosting!
At this point though, Sourceforge is probably thinking that if they even charge a fee as low as $10/month they'd lose all the perty market share. I disagree, if for the unique services they provide alone. In order for this to play, those services and bandwidth need to keep themselves to very high standards.
I just know someone is going to find something wrong with my math
The line must be drawn here. This far. No further.
On Sunday I saw a news item on SF regarding a transition to Oracle. Yesterday it was gone. How convinient. You'll be hard-pressed to find source code for SourceForge on SourceForge. For *any* part of it. Not long ago you could visit SF's project page and browse the CVS. It's now gone and all file releases have been wiped out. The excuse up to this point has been that there are proprietary extensions and source for those isn't available. Ok, where's the rest then? This looks bad. Back when ESR came up defending VA's position I was skeptic (read my diary). Now I'm convinced VA's going the wrong way. I'm moving my code out of SF and I'll feel uncomfortable contributing to projects hosted on SF.