Two Microsoft licenses submitted for OSI approval [LWN.net]

Two Microsoft licenses submitted for OSI approval

Posted Aug 13, 2007 17:43 UTC (Mon) by edschofield (guest, #39993) [Link] (17 responses)

I'm amazed. Could this really be a pro-open-source stance from Microsoft?

Two Microsoft licenses submitted for OSI approval

Posted Aug 13, 2007 17:52 UTC (Mon) by fandom (subscriber, #4028) [Link]

No.

Releasing significant software under these licenses might though

Microsoft's position isn't anti-open-source, it's against their competitors (esp. GPL)

Posted Aug 13, 2007 18:06 UTC (Mon) by dwheeler (guest, #1216) [Link] (15 responses)

Microsoft is NOT anti-open-source, and hasn't been for years. Key programs like Windows and Office include open source code (e.g., some TCP/IP utilities from BSD, zip code, etc.). They've even created some open source software programs (e.g., IronPython, WiX), and manage an open source site (CodePlex).

Microsoft loves open source that helps them develop or sell their products. So you can't say in a blanket way "Microsoft opposes open source software". It's not that general.

What they don't like are open source programs that compete with their products, or strategies that make it impossible for them to extinguish the competition. So they'll help Python... but work hard against the Linux kernel or OpenOffice.org. They love BSD licenses... but work hard against the GPL. The GPL, in particular, counters the embrace, extend and extinguish approach very effectively - so Microsoft fights hard against it.

I believe different leaders within Microsoft have very different ideas about what Microsoft should do about open source software. That would explain their self-conflicting responses: they're not sure what to do, either.

Microsoft's position isn't anti-open-source, it's against their competitors (esp. GPL)

Posted Aug 13, 2007 20:18 UTC (Mon) by epa (subscriber, #39769) [Link] (9 responses)

Yes Windows includes the TCP/IP stack taken from BSD, but it is rather a stretch to call it open source code. Microsoft doesn't include the source code and doesn't provide any way for you to change the code and run your own changed version, let alone share it with others. An unchangeable binary blob doesn't really fit the open source definition, whatever the evolutionary origin of the code.

OTOH, as you say, Microsoft has made true open source software. On the other hand, they have taken deliberate measures to hamper open source, such as claiming to document information about Windows network protocols (to satisfy European Union antitrust regulators) while putting it behind a licence agreement that specifically excluded GPLed implementations.

Microsoft's position isn't anti-open-source, it's against their competitors (esp. GPL)

Posted Aug 14, 2007 0:51 UTC (Tue) by brouhaha (subscriber, #1698) [Link] (2 responses)

Yes Windows includes the TCP/IP stack taken from BSD, but it is rather a stretch to call it open source code.

Au contraire, that's a perfect example of the difference between "Open Source" and "Free Software". Open Source allows users to do almost anything they like with it, other than removing copyright notices. That includes turning it proprietary. Thus Microsoft's use of BSD code is fully within the Open Source paradigm. The original BSD code is still Open Source, but Microsoft's derived code is not.

Free Software, on the other hand, is based on licenses that require the preservation of freedom, and don't allow third parties to turn it proprietary.

Microsoft's position isn't anti-open-source, it's against their competitors (esp. GPL)

Posted Aug 14, 2007 2:23 UTC (Tue) by josh (subscriber, #17465) [Link]

Those terms do not mean what you think they mean.

Both copyleft licenses (such as the GPL) and non-copyleft licenses (such as the BSD licenses) qualify as both Open Source and Free Software.

Totally untrue

Posted Aug 14, 2007 2:28 UTC (Tue) by stevenj (guest, #421) [Link]

Open Source allows users to do almost anything they like with it, other than removing copyright notices. [...] Free Software, on the other hand, is based on licenses that require the preservation of freedom, and don't allow third parties to turn it proprietary.

Um, no. This is the different between a copyleft license (e.g. the GPL) and a non-copyleft permissive license (e.g. the X11 license). Software under both types of license are considered free software and open source by the Free Software Foundation and the Open Source Initiative, respectively.

Microsoft's position isn't anti-open-source, it's against their competitors (esp. GPL)

Posted Aug 15, 2007 14:15 UTC (Wed) by gdt (subscriber, #6284) [Link] (5 responses)

Yes Windows includes the TCP/IP stack taken from BSD...

For the record, Microsoft say that this has not been true since 1994. By my count Microsoft claim to have completely re-written the TCP/IP implementation in its Windows operating systems three times since the original third-party but BSD-derived implementations in WfW and NT 3.1: NT 3.5 (NT4, 95, 98, Me), 2000 (Xp, 2003), Vista.

"Completely" might be overstating things, since 2000 had some bugs in its design choices which were shared by earlier BSDs. The implementation in Vista appears to be very different to previous implementations (and much better too).

Microsoft's position isn't anti-open-source, it's against their competitors (esp. GPL)

Posted Aug 16, 2007 9:03 UTC (Thu) by tialaramex (subscriber, #21167) [Link] (4 responses)

Yeah, this particular mis-information really needs to go away. Someone ought to compile a generic list of IT forum myths so that every time a newbie says "I heard..." or "Isn't it true..." they can get a one day time out and a pointer to the list of myths. Not so much here on LWN but in general technology forums where the same myths come up over and over.

* NO Windows doesn't include a BSD-derived TCP/IP stack and NO they didn't just snip the copyright messages off. NO the BONE TCP/IP stack that was never officially shipped yet somehow seems to be on a lot of BeOS systems isn't BSD-derived either, and NO that wouldn't make it legal to copy it anyway.

* NO the X window system doesn't send everything via TCP/IP networking when you run software locally. NO other systems don't, on the whole, build the GUI into the OS kernel, and NO X wouldn't go "a lot faster" if you ripped the support for network transparency out.

* NO Bill Gates never said that about 640K of memory. NO it's not in "some book" you read unless it's a misquote in that book too. NO it isn't clever to just make up quotes in order to poke fun at someone you don't like.

* NO Linux did not start out as a "server OS" unless suddenly student university digs are server rooms and reading Usenet is a server activity. NO it wasn't "designed for" servers either, except in the sense that your servers happen to be i386 hardware, much like my desktop.

I'm sure there are dozens more, at least...

Anti-Myths: Actually, Windows DOES use some BSD code

Posted Aug 16, 2007 17:20 UTC (Thu) by dwheeler (guest, #1216) [Link] (1 responses)

"NO Windows doesn't include a BSD-derived TCP/IP stack and NO they didn't just snip the copyright messages off."

Well, that depends on the definition of "stack". There's BSD-derived code in Windows for implementing TCP/IP. A trivial hunt on Windows XP with "strings" showed that \WINDOWS\System32\nslookup.exe includes "Berkeley" (I'm sure there's more, but I only need one example for the point). And this is OKAY; the Berkeley licenses explicitly PERMIT reuse.

Anti-Myths: Actually, Windows DOES use some BSD code

Posted Aug 29, 2007 19:34 UTC (Wed) by tialaramex (subscriber, #21167) [Link]

The only BSD-derived code is in assorted utilities and examples like this. Most of them are obsolete (nslookup is considered broken by design, FTP is hopelessly insecure, as is telnet) and are included because they're small and Microsoft has learned that removing even the most insignificant feature from their products causes people to accuse them of "downgrading".

Note that I didn't write "network stack" or "operating system stack" which might arguably include such independent programs, but only the TCP/IP stack. Maybe that's nitpicking.

Microsoft's position isn't anti-open-source, it's against their competitors (esp. GPL)

Posted Aug 16, 2007 19:19 UTC (Thu) by bronson (subscriber, #4806) [Link] (1 responses)

> NO the BONE TCP/IP stack that was never officially shipped yet somehow seems to be on a lot of BeOS systems isn't BSD-derived either

Cite your source? The BeOS networking stack most certainly WAS derived from BSD. I have the evidence sitting on an old dual 603e BeBox in my garage.

Microsoft's position isn't anti-open-source, it's against their competitors (esp. GPL)

Posted Aug 29, 2007 19:48 UTC (Wed) by tialaramex (subscriber, #21167) [Link]

Given that you don't even seem to understand that Be Inc. had two different TCP/IP stacks, I doubt that you have any interesting "evidence".

Howard Berkey, who actually worked on the project, said several times that the BONE stack was not BSD-derived. If he lied then Be's successors in interest stand to be sued for copyright infringement for stripping the BSD copyright notices. The authors would certainly be interested in your evidence, if in fact it exists.

As to the net_server stack, I don't think anyone ever cared how Be Inc developed that, whatever they did was mind-blowingly stupid, its performance is wretched compared even to other userspace TCP/IP offerings.

Microsoft's position isn't anti-open-source, it's against their competitors (esp. GPL)

Posted Aug 13, 2007 21:04 UTC (Mon) by okeydoke (guest, #46751) [Link] (4 responses)

"""They've even created some open source software programs (e.g., IronPython, WiX)"""

No, Microsoft've not created IronPyton, Microsoft bought it.

IronPython

Posted Aug 14, 2007 4:42 UTC (Tue) by sanxiyn (guest, #44599) [Link] (3 responses)

"No, Microsoft've not created IronPyton, Microsoft bought it."

Microsoft bought IronPython 0.6, developed it further for more than 3 years, and released IronPython 1.0 and subsequent versions. In my book that qualifies as "creating IronPython".

IronPython

Posted Aug 14, 2007 9:24 UTC (Tue) by arcticwolf (guest, #8341) [Link] (2 responses)

"Microsoft bought IronPython 0.6 [...] In my book that qualifies as 'creating IronPython'."

Y'know, unless this was intended to be ironic, you may have a career as Rove's successor if you can make claims like that with a straight face.

IronPython

Posted Aug 14, 2007 9:50 UTC (Tue) by sanxiyn (guest, #44599) [Link] (1 responses)

It wasn't intended to be ironic. How doesn't 3 years worth of development qualify as "creating some open source software programs"?

Let's review:

A: Microsoft created some open source software programs like IronPython.
B: No, Microsoft bought IronPython.
C: Microsoft created IronPython in the sense that they developed it further after buying.
D: Developing further is not creating. Are you being ironic?
C: ???

Of course developing further is not creating, but I hope the context is obvious (creating some open source code) to any readers.

IronPython

Posted Aug 16, 2007 8:46 UTC (Thu) by ekj (guest, #1524) [Link]

Because people generally consider "create" as in made it come into being. So it's strange to say someone *created* something when that something existed before they even touched it. If you'd written: They made major *improvements* to IronPython. We'd have agreed. (well, atleast I would've)

What is the difference between the two licences?

Posted Aug 13, 2007 20:58 UTC (Mon) by epa (subscriber, #39769) [Link] (3 responses)

At first glance it appears that the 'community' licence is copyleft and 'permissive' is Apache-style. But the only difference in the licence text appears to be this clause:

>(A) Reciprocal Grants- For any file you distribute that contains code from
>the software (in source code or binary format), you must provide recipients
>the source code to that file along with a copy of this license,
>which license will govern that file. You may license other files that are
>entirely your own work and do not contain code from the software under any
>terms you choose.

This appears in the community licence only. However, both licences say:

>If you distribute any portion of the software in source code form, you
>may do so only under this license by including a complete copy of this
>license with your distribution.

And the introductory text also says that you may not redistribute MS-PL code under a different licence; only under the MS-PL. So it seems that both licences are equally 'viral', and in practice, identical. The only difference may be that the MS-PL allows you to distribute binaries under your own licence, but only if you don't include source code, which would be a strange requirement coming from anyone except Microsoft.

Also it is not clear what the legal meaning of a 'file' is. What about if you distribute a single tar file containing source code?

What is the difference between the two licences?

Posted Aug 14, 2007 4:27 UTC (Tue) by jamesh (guest, #1159) [Link] (2 responses)

That last clause doesn't seem any more viral than the BSD license. If you distribute BSD source code, you are bound to do so in compliance with the BSD license. The same appears to be true for the MS-PL.

In both cases the license does not claim to apply to other code that you combine with it, so it is hard to see how you could claim that it is viral.

What is the difference between the two licences?

Posted Aug 16, 2007 12:37 UTC (Thu) by epa (subscriber, #39769) [Link] (1 responses)

>If you distribute BSD source code, you are bound to do so in compliance
>with the BSD license.

In compliance with the BSD licence, but not necessarily under the BSD licence. You are allowed to take BSD source code and distribute it under a different licence if you wish. As far as I can tell, MS-PL source code can be distributed only under the MS-PL. So it seems to be a kind of weak copyleft, not as 'viral' as the GPL (actually, that's a silly word, and hardly any more accurate than 'cancer'), but perhaps comparable with the LGPL, depending on the legal definition of 'files'.

What is the difference between the two licences?

Posted Aug 17, 2007 15:37 UTC (Fri) by jamesh (guest, #1159) [Link]

If I release some code under the BSD license and you then combine it with your own code, you can release the combination under pretty much any terms you want.

If you distribute the source code to someone, and that person extracts the portion that was under the BSD license, then they can use it under those terms (you did leave the copyright and permission notice in place like the license requires, right?).

This is not the same as conversion of LGPL code to GPL, which is an option given by the original author to people wanting to redistribute the code.

From my reading, the MS-PL clause is effectively equivalent to BSD here (even if the wording is slightly different).

Two Microsoft licenses submitted for OSI approval

Posted Aug 14, 2007 9:30 UTC (Tue) by etienne_lorrain@yahoo.fr (guest, #38022) [Link]

Maybe not the right place to ask, but:
Licensing, is it a form of selling or a form of renting?
If one's break the terms of the license, the product is no more yours, you just own the support - so is that renting?
The free licenses usually talk of "distributing" or equivalent, but there isn't any distribution if renting (if you rent a car with a GPL based audio system, you do not have the right to the source - you need to own the car for that).
Where didn't I get it?
Did I rent-for-life or buy my Windows License?

Two Microsoft licenses submitted for OSI approval

Posted Aug 14, 2007 12:17 UTC (Tue) by dag- (guest, #30207) [Link] (2 responses)

I see 2 problems with these new licenses.

1. What do these licenses offer over existing licenses ? Does those differences validate a new OSI approved license ? My concern here is that having a license for every different characteristic makes only so much sense until we end up with thousands of OSI licenses, for every company one or more.

2. Having 'Microsoft' in the license name seems to be strange. Why brand a license with your company name in the first place ? Isn't Open Source meant to be a place where everyone is equal ? Sure, this is not the first license that carries a company or product name, so I'm not accusing Microsoft of dirty practices (not this time anyway). Still, I see no clear benefit to having Microsoft in the license name. Only liabilities and wrong connotations as something coming from Microsoft, while the software is definitely not.

I do think OSI is responsible for making sure that every license being approved offers something unique and is company-agnostic. In name and content.

Otherwise I'll duplicate the GPL and call it Dag's Own Public License and go for OSI validation to generate some publicity and use the OSI licensing process as a marketing tool.

I hope someone from OSI is paying attention and not setting any bad examples.

Two Microsoft licenses submitted for OSI approval

Posted Aug 14, 2007 15:15 UTC (Tue) by hingo (guest, #14792) [Link] (1 responses)

While I agree that all of your points are very good... Have you looked at the licenses OSI have already approved? The number of them? Their names?

Two Microsoft licenses submitted for OSI approval

Posted Aug 16, 2007 10:05 UTC (Thu) by MathFox (guest, #6104) [Link]

IIRC the OSI has changed its policies on license acceptance somewhat. Aspects like "is the license sufficiently different from other licenses" are taken into account now and one looks at the branding of the license. It is to avoid triplicates like NPL, MPL and CDDL. I think that the two Microsoft submitted licenses are sufficiently different to enter the evaluation process; I am pretty sure that the naming issue will come up there too.

Microsoft has made source code available over the years, under various licenses. While Microsoft certainly isn't an Open Source company, they created, used and distributed some Open Source software. We should praise them for what they do good and criticize them (fairly) for what they do wrong. Have the OSI decide whether these Microsoft licenses fit the Open Source Definition and check that they do so thoroughly and fairly.

IANAL, not speaking for OSI either.