llvm dev - May 2008 - [LLVMdev] Forward: Discussion about custom memory allocators for STL

On Sat, May 17, 2008 at 9:28 PM, Dominic Hamon <dom.hamon at gmail.com>
wrote:
> There is a thread elsewhere on this mailing list illustrating how
> important it is for the maintainers of LLVM to keep LLVM usable in a
> commercial environment. As such, I would strongly recommend avoiding
> Boost as it has a bad name in some quarters, regardless of its license,
> for including work that is not safe for commercial users to take on. Ie,
> there are so many contributors, and their contribution tracking has been
> poor in the past, that business affairs departments in commercial
> companies, and their associated patent lawyers, are unable to determine
> how much of Boost is truly the authors' own work, and how much is
borrowed.
>
> Given the delicate relationship between the commercial sector and open
> source, adding Boost usage to LLVM will harm the commercial sectors view
> of this product.
>
Do you have a source for that opinion?  I've never heard that view of Boost.

On Sat, May 17, 2008 at 7:41 PM, me22 <me22.ca at gmail.com>
wrote:
> On Sat, May 17, 2008 at 9:28 PM, Dominic Hamon <dom.hamon at
gmail.com> wrote:
>> There is a thread elsewhere on this mailing list illustrating how
>> important it is for the maintainers of LLVM to keep LLVM usable in a
>> commercial environment. As such, I would strongly recommend avoiding
>> Boost as it has a bad name in some quarters, regardless of its license,
>> for including work that is not safe for commercial users to take on.
Ie,
>> there are so many contributors, and their contribution tracking has
been
>> poor in the past, that business affairs departments in commercial
>> companies, and their associated patent lawyers, are unable to determine
>> how much of Boost is truly the authors' own work, and how much is
borrowed.
>>
>> Given the delicate relationship between the commercial sector and open
>> source, adding Boost usage to LLVM will harm the commercial sectors
view
>> of this product.
>>
>
> Do you have a source for that opinion?  I've never heard that view of
Boost.
I would also like to see some data on that.  I have heard it once or
twice before on mailing lists, though never any evidence to back it
up, and never from within an actual business.

-- 
Cory Nelson

me22 wrote:
> On Sat, May 17, 2008 at 9:28 PM, Dominic Hamon <dom.hamon at
gmail.com> wrote:
>   
>> There is a thread elsewhere on this mailing list illustrating how
>> important it is for the maintainers of LLVM to keep LLVM usable in a
>> commercial environment. As such, I would strongly recommend avoiding
>> Boost as it has a bad name in some quarters, regardless of its license,
>> for including work that is not safe for commercial users to take on.
Ie,
>> there are so many contributors, and their contribution tracking has
been
>> poor in the past, that business affairs departments in commercial
>> companies, and their associated patent lawyers, are unable to determine
>> how much of Boost is truly the authors' own work, and how much is
borrowed.
>>
>> Given the delicate relationship between the commercial sector and open
>> source, adding Boost usage to LLVM will harm the commercial sectors
view
>> of this product.
>>
>>     
>
> Do you have a source for that opinion?  I've never heard that view of
Boost.
>
>   
The company that I work for that had to pull boost out of a project 
after the patent lawyers couldn't reliably determine the sources for 
some of the components of boost.

There are other arguments elsewhere on this mailing list against using 
boost that are far more specific, however this does come from personal 
experience and I thought it important to share.

It has also been mentioned that it is also true for LLVM itself, and it 
is. However, the amount of code in LLVM, and the specific nature of its 
usage and what it is for lends itself to better contribution tracking 
immediately. Boost is a nebulous piece of software with a wide range of 
functionality.

Dominic

On Sun, May 18, 2008 at 7:59 PM, Dominic Hamon <dom.hamon at gmail.com>
wrote:
> It has also been mentioned that it is also true for LLVM itself, and it
> is. However, the amount of code in LLVM, and the specific nature of its
> usage and what it is for lends itself to better contribution tracking
> immediately. Boost is a nebulous piece of software with a wide range of
> functionality.
>
If I were worrying about patents, though, I'd be much more worried
about LLVM than Boost.  Much of Boost is original, and could well be
prior art against patent claims, whereas LLVM could easily use a
patented register allocator or similar without anyone knowing.