kern/29355: [kernel] [patch] add lchflags support

[Josh Goodall]

On 14/11/2012, at 8:11 AM, Mark Linimon <linimon at lonesome.com> wrote:

> On Wed, Nov 14, 2012 at 08:07:02AM +1100, Josh Goodall wrote:
>> The handling of this PR is the reason I stopped contributing to FreeBSD.
> 
> We simply get more PRs than we can handle.  I know that's not an
> answer that will satisfy anyone, but it's the truth.
> 
> mcl

I filed this PR over a decade ago, so really "we didn't get around to it" seems like a thin excuse.

But there's more to it than that. When originally lodged, I'd been interested in FreeBSD for a while, contributed several ports and the odd userland PR, I'd spent a few months learning the kernel structure, looking for a small real-world project to get started.

I liked hacking on FreeBSD a lot; I had hopes of eventually earning a commit bit. I did a very thorough job on the lchflags patch, finding every possible userland tool (cp, rm, tape, chflags, mtree etc etc) that needed to learn about it, and produced a patch that bright (or maybe cmc) described (very memorably, to me, the hopeful novice) on IRC as "a textbook example of adding a system call". Man was I proud of that review.

Still, no-one wanted to actually commit it. I figured, oh well, eventually I'll find a taker. I updated it every now and then to ensure currency.

Then, one day in 2002, mux committed a half-assed lchflags that didn't have much of the userland stuff, contained a syscall signature inconsistent of the rest of the VOP_SETATTR calls (making base system mods much more complicated), and sans credit.

No-one seemed to care, and, having had my novice work so brusquely sidelined, I lost interest in contributing further to FreeBSD and moved on to other projects.

This is all water long under the bridge, of course, but I never went on the record about it before. Take it as a parable of how to lose a contributor, if you like.


Josh.