LWN: Comments on "Brazilian government signs up to develop OpenOffice and LibreOffice (The H)"

Brazilian government signs up to develop OpenOffice and LibreOffice (The H)

intgr — Mon, 11 Jul 2011 19:17:28 +0000

> The fact that major forks of major projects tend not to go anywhere in the
> long run is my point.

But that doesn't make forks irrelevant. I think these (short-term) forks are still potentially harmful. They're taking the work of the community and making money off it without giving anything back. They're even competing with the original community product.

With the GPL, such a business model would be harder. Either these forks wouldn't exist in the first place, or they would end up benefiting the community as well.

As for BSD OS, isn't Darwin a full-blown and fork maintained by a company?

Brazilian government signs up to develop OpenOffice and LibreOffice (The H)

kov — Mon, 11 Jul 2011 04:22:14 +0000

You're being too optimistic. I'll be happy to see even 1 patch being contributed. If that ever happens it'll be time to worry about which one they'll drop. =(

Brazilian government signs up to develop OpenOffice and LibreOffice (The H)

butlerm — Sun, 10 Jul 2011 16:43:17 +0000

The issue is not a proprietary fork so much as a proprietary fork where the code base diverges so much that developers no longer contribute to the original code base. Of the PostgreSQL forks you have mentioned, I would be curious to know what degree that is the case.

Illustra is more or less dead. Acquired by Informix, and Informix by IBM, it is now little more than a historical footnote as far as I can tell. The story with most BSD forks is similar. The fact that major forks of major projects tend not to go anywhere in the long run is my point. It doesn't matter whether the license is BSD or GPL because the economics of maintaining a completely separate code base for major projects is cost prohibitive.

Suppose the Linux kernel and all its major components were BSD licensed. We might have Oracle Linux, HP Linux, IBM Linux all tweaky and somewhat more incompatible than different Linux distributions are today. Given the lesson of the past thirty years, however, would any of those three firms fund completely diverged code bases all over again? Even with the freedom to do so, the idea that any of those companies would maintain completely diverged, non-narrow market forks of the Linux kernel over the long haul seems pretty dubious to me. It just doesn't work, and the GPL is not the primary reason why that is the case. It is arguably almost irrelevant.

Brazilian government signs up to develop OpenOffice and LibreOffice (The H)

intgr — Sun, 10 Jul 2011 11:27:16 +0000

> So why haven't they forked PostgreSQL (completely) and made a derivative
> with all sorts of competitive advantages that they don't share with anyone
> else?

Just because you haven't heard of them doesn't mean they don't exist. There are several proprietary PostgreSQL forks out there, but they're not advertised as forks. Among others: Greenplum, Netezza, Illustra, Yahoo's distributed database, Vertica (suspiciously similar to Postgres, but supposedly doesn't share any code)

EnterpriseDB is different, they use open source as a selling point. Most of their customers are jumping ship from Oracle because they want to free themselves of vendor lock-in.

Brazilian government signs up to develop OpenOffice and LibreOffice (The H)

rahvin — Sun, 10 Jul 2011 09:59:42 +0000

And the public statements by IBM and the history of Unix indicate that BSD licensing means companies aren't willing to contribute back if others can use and improve without returning the favor. For those that forget, IBM when they started with Linux made it clear that it was only because of the GPL that they would contribute all their major innovations into Linux because the GPL guaranteed they could use any improvements that anyone made to the code.

You see the same with the Unix forks, they all had access to the same code base and then diverged the code in incompatible directions to create customer lock in. And as you noted the same thing happened with the BSD's themselves in that they diverged the forks to the point that the code was incompatible.

There are successful BSD projects, personally I think they are only successful because they got there first. By that I mean the project was developed and established before there was a serious GPL project competitor. Why create a GPL apache when the existing one is already so good and has so many deployments? Once a software package is established (lots of deployments, support and things like plugins are developed) it's hard to displace it regardless of quality or features.

Personally I'm very confident that Openoffice and Libreoffice will diverge very quickly and that LO will be the winner. There are already so many developers behind LO that I just don't see how the BSD version can succeed without several companies paying for lots and lots of developers. Something I doubt will happen. In fact I have no doubt in my mind that at least in IBM's case they won't contribute back anything that improves Symphony.

Brazilian government signs up to develop OpenOffice and LibreOffice (The H)

dlang — Sun, 10 Jul 2011 05:51:48 +0000

the various proprietary forks of BSD lead to the opposite conclusion

Brazilian government signs up to develop OpenOffice and LibreOffice (The H)

butlerm — Sun, 10 Jul 2011 00:51:48 +0000

This is an excellent test of the relative merits of BSD style vs. GPL style licensing in large opensource projects. If IBM, Oracle, et al contribute enough to OpenOffice it might outpace LibreOffice, making an implicit case that open BSD style licensing is worth the cost. Or the reverse could be the case, and those who don't want anything to do with a project with proprietary forks could leave OpenOffice in the dust.

The best argument I know for BSD style licensing is that most proprietary forks are relatively short lived, and worth the cost benefit trade off of attracting additional multi-vendor contributions. Short lived, because the companies who can bear the cost of maintaining a full blown fork of a large, healthy open source project are few, and most don't want to do it anyway. The proprietary forks of PostgreSQL are an excellent example.

HP needs a database to compete with Oracle and IBM, for example. So why haven't they forked PostgreSQL (completely) and made a derivative with all sorts of competitive advantages that they don't share with anyone else? EnterpriseDB has a derivative with some extra features, but they don't fork the base code either. It is my understanding that they contribute quite a bit in the reverse direction.

The practicalities of maintaining a completely independent fork all by yourself make BSD licensed projects much more successful than one might expect. Proprietary forks make a little revenue for someone, and die. And if the contributions in return outweigh the cost, it seems like a pretty good trade-off to me.

Brazilian government signs up to develop OpenOffice and LibreOffice (The H)

Hausvib6 — Fri, 08 Jul 2011 02:12:27 +0000

Let's which one is dropped once the Brazillian get tired of contributing to two possibly-diverging codebases.