incubator-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gregory Nutt <spudan...@gmail.com>
Subject Re: [DISCUSS] NuttX Proposal
Date Wed, 27 Nov 2019 00:49:24 GMT
Hi, Justin,

It is good to meet you.  Thank you for taking the time to respond.

- Apache has a no dictator rule (friendly or otherwise), given there’s a 
person who may be in that role currently in the project, are they fine 
with giving up control to the rest of the community?

Yes, I am fine with giving up control.  I have been doing this for a 
very long time (probably daily for 15 years) and I am perfectly happy 
with others taking the lead.  So far all of the people who have been 
interested are of the highest technical caliber and I have complete 
confidence that they can share the leadership roles.

- The contributors may need to sign ICLAs. Given the number of people 
involved and the age of the project, if this ia required, it might be 
difficult.

Some of the files are so old that it may not be possible to even contact 
all of the authors.  I am the committer on all files and also listed as 
the author in the CM systems up to a few years ago (other people are 
given credit only in ChangeLogs and ReleaseNotes). Most have given 
copyright rights to me.  Is it the author, the committer, or the 
copyright holder that matters here?

- Given the large number of committers, I expected the initial committer 
list to be much larger, have you considered inviting all of the major 
committers?

No one but I can commit to the existing repositories so I am the 
committer of everything.  The names are not actually committers, but 
authors from the git commit --author= argument.  In the early years CVS 
and SVN were used and in the early GIT years, I did not yet know about 
the --author= option.  So you can clearly see when I discovered 
--author= in the data at: 
https://bitbucket.org/nuttx/nuttx/addon/bitbucket-graphs/graphs-repo-page#!graph=contributors&uuid=d90bf438-7869-4921-8926-fd25b0901043&type=c&group=weeks

Maybe around 2015?

- What does the community as a whole think abut this move? I can see 
this here [1], but I was a little surprised there wasn’t more public 
discussion about this. Is there consensus that they want to join the ASF 
and understand what that means?

I have only heard positive things lately.  There was some grumbling 
about changing the licenses some time back, but I think we are past 
that. Many people had the mistaken impression that the Apache license in 
some way put more burden on the end-user.  The complexity of the 
language used in the license is a barrier to many people.

- Is the community aware of some of the issues they may encounter in 
moving to ASF infrastructure and doing things in an Apache Way? (release 
process usually being the main difference).

They are aware superficially but I don't think most appreciate the full 
implications.

If there were not a full PMC supporting the project then it would be a 
problem for me as well.  With the PMC we should be able to share the 
load.  I have done 133 releases of NuttX alone.  I am hoping to be 
replaced with a more Apache-savvy release manager.  I would prefer to be 
this assistant, advisor / helper / mentor on most of the PMC issues.

- I notice you mention BSD-4 cause in you list of licenses, this is 
actually Category X and not compatible with the ALv2 license. However 
if  it is what I think it is, it easy to work around (as we did for 
Mynewt).

There are a few, very old 4-clause BSD files in the C library (leveraged 
from old, old BSD C library files).  Don't copyrights have a shelf life 
like patents?  I am sure that these are too old to be an issue because 
the copyrights have expired anyway.  This should be all of them:

    $ find . -name *.c | xargs grep " California, Berkeley"
    ./fs/nfs/rpc_clnt.c: * University of California, Berkeley and its
    contributors. 4. Neither the
    ./libs/libc/stdio/legacy_dtoa.c: *      California, Berkeley and its
    contributors.
    ./libs/libc/stdio/lib_libdtoa.c: *      California, Berkeley and its
    contributors.
    ./libs/libc/stdlib/lib_bsearch.c: *      California, Berkeley and
    its contributors.
    ./libs/libc/stdlib/lib_qsort.c: *    California, Berkeley and its
    contributors.

    $ find . -name *.h | xargs grep " California, Berkeley"
    ./fs/nfs/rpc.h: *      California, Berkeley and its contributors.

Six files.  That is basically just the RPC logic in NFS client and some 
pieces of the C library.  bsearch() and qsort() probably have 
replacements with other licenses.  dtoa() probably does not.

If the proposal does come to the Incubator, (and the project thinks I'm 
a good fit), I can help out as a mentor. I‘m mentored a number of IoT 
projects here at Apache, including Mynewt.

That is very good news!  Thank for that.

I spoke with the Mynewt folk a couple of years back about incorporating 
some their IoT components into NuttX a few years back.  Licensing was an 
obstacle then, but times are changing.

Greg



Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message