Subject: RE: libpthread
To: Frank van der Linden <>
From: Sporleder, Matthew (CCI-Atlanta) <>
List: tech-kern
Date: 06/20/2003 09:54:15
Speaking of threads.
I was under the impression that this was our main stopping point for =
Are we approaching native java support?  That would be -very- good.

-----Original Message-----
From: Frank van der Linden []
Sent: Friday, June 20, 2003 8:57 AM
Subject: Re: libpthread

On Fri, Jun 20, 2003 at 03:52:17PM +0300, Mihai CHELARU wrote:
> Don't think it's such a good idea. Most of the users will go for the=20
> stable solution and the bugs in SA will be detected slowly. After all, =

> -current is not a stable release and everybody should use it on his =
> risk. The only solution that I see is to see that more developers are=20
> becoming familiar with SA code. As far as I read, only nathanw is able =

> to code this side of the kernel at this moment. And, as we all know,=20
> it's not good for an entire project to depend on one single person.

Yes, this (and Dan's comment) are the core of the issue. First of all,
if it hadn't been brought into -current, it would have evolved
slowly on a branch, with only Nathan working on it, and the burden
of doing all the branch syncs (they cost a considerable amount of time)
by himself. There was no alternative then to bring it into -current.

I agree that it would have been a good idea to simply disable
SAs for MP kernels, but other than that, this was the only way
to go forward. Anyone who can do better is free to take my
place in Core anytime.

And you're absolutely right, it's bad to have only one person really
understand the details of such a major subsystem. Which is why everyone
is encouraged to look at it and contribute. Fortunately, one person
has stepped forward and will become a NetBSD developer so that he
can contribute directly to the SA code.

- Frank

Frank van der Linden                                            =
NetBSD. Free, Unix-like OS. > 45 different platforms.    =