ietf
[Top] [All Lists]

Re: bettering open source involvement

2016-08-03 04:38:31
On Wed, Aug 3, 2016 at 11:10 AM, Philip Homburg
<pch-ietf-5(_at_)u-1(_dot_)phicoh(_dot_)com> wrote:
There are. But if the IETF wants to encourage code that anybody can study
and/or borrow, GPL is not the way to go.

I don't think the "study" argument flies anymore. It's akin to banning
a doctor from reading "one flew over the cookoos nest".

Or a barber from watching "sweeny todd".


With all the discussion about code, I'm much more interested in good APIs.

I think the advanced socket API for IPv6 is one that works really well.

I am still in search of common, simple means, to access and change the
ecn related bits in the IP header, particularly for udp packets. The
"standard" for doing such is varied, actually getting at the bits per
packet (for OSes that can do it) requires the cmsg facility.

More recently, getdns defines for the first time a sensible API for DNS.
(A bit verbose if you write C, but certainly at the right level of
abstraction).

That is awesome!

https://getdnsapi.net/spec.html

Thank you for bringing it to my attention. Their proposed standard has
a few problems - no mdns support for one - and a codebase issue
(openssl dependency primarily), but it is otherwise looking like a
vast improvement on older dns apis.

For APIs, you'd like to have at least two compatible implementations of a
single API. So if one of those is GPL and the other BSD, there is no
licensing issue.

There may be other uses for code, for example to illustrate a complex
algorithm. But the one thing I miss most are well defined APIs.

+10






-- 
Dave Täht
Let's go make home routers and wifi faster! With better software!
http://blog.cerowrt.org