Haskell Platform Proposal: HLint

[Simon Marlow]

On 12/11/2010 16:48, Duncan Coutts wrote:
> On 12 November 2010 12:56, Simon Marlow<marlowsd at gmail.com>  wrote:
>> On 12/11/2010 12:12, Duncan Coutts wrote:
>>
>>> Secondly, yes I agree that people tend not to care too deeply about
>>> this. To play devil's advocate, if people don't really care then
>>> instead of saying that means we must have an exception, perhaps that
>>> means that we're ok to comply with the letter of the law but not the
>>> spirit. It is a big hassle to work out wording for an exception and
>>> get all license holders of each project to agree (I did it once with
>>> Gtk2Hs and LGPL-2.0->2.1). If people do not actually care, why bother?
>>
>> Because while the authors might not care, potential users of the library
>> might well care a lot.  Some of us have to get licenses past corporate
>> lawyers, and they generally don't even agree about what the LGPL means.  The
>> more of the LGPL we can exclude, the easier this process might be.
>>
>> Avoiding all this nonsense entirely is for some people by far the easiest
>> option, though of course I recognise the right of authors to choose whatever
>> licensing terms they want for their work.  I mainly want to ensure that
>> there is a meaningful option for users who want or need to avoid the
>> GPL/LGPL for whatever reasons.
>
> Presumably you mean that users might need to re-review the LGPL in how
> it applies to Haskell code when compiled by GHC, since we have
> afterall been shipping GPL tools and LGPL C libs along with ghc for
> some years.

Tools are somewhat easier, and we have a way to avoid the LGPL lib (GMP) 
if necessary.  Also, policies change over time, and new license 
dependencies are often subject to more scrutinee than those for software 
that is already being used.

> So the theory is that the corporate lawyers might re-OK it
> for Haskell code if it's clear that the only place where Haskell might
> be different there is an exemption anyway. On the other hand there's
> the danger they'll grumble even more about license extensions written
> by non-lawyers.

This is true, yes - non-standard licenses might make more work, not 
less.  Usual NAL disclaimers apply, but I imagine that an exception of 
the form "ignore paragraphs X--Y" ought to be unambiguous and might make 
the license/software usable for some people where it wasn't previously.

> I know it's not trivial getting any open source code past corporate
> layers. I've had to go through the process myself (BSD and LGPL C and
> Haskell code). I've noted that our tools are not as good as they could
> be, e.g. in collecting all the LICENSE files of all dependencies (I've
> had to do it by hand). I'm sure that lots of us are regularly
> violating the letter of the BSD license. Galois have been working on
> some features to add to Cabal to help us all automate some of this
> (we've done one round of patch-review so far).

Yes, I like the sound of that.

Cheers,
	Simon