Garrett,

Bluntly, this is a poor solution to a non-problem -- and I'm dismayed that our small, pragmatic, highly technical community has been inflicted with what is effectively a thermonuclear troll. Please, for the sake of the community: stop.

If you want to be inspired by the work done by ZFS and DTrace and invent some new, orthogonal mechanism for an illumos-derived OS to identify its features, fine; we can get consensus on that and get that integrated -- but uname itself should be left alone, for all of the reasons outlined by Jonathan, Hans and others. Which is to say: a more iterative approach needs to be taken here, starting with the things everyone agrees on (like illumos cpp defines).

To sum, we are far too small (and have far too many hard problems to solve) to waste our time like this; let's please get back to work on the meaningful technical problems in front of us (e.g., lx branded zones). Speaking for myself personally: I will not be engaging in any further e-mail discussion on this proposal -- and my vote on the Dev Council for this proposal would be/will be an emphatic "no."

- Bryan

On Mon, May 12, 2014 at 8:36 PM, Garrett D'Amore via illumos-developer <developer@lists.illumos.org> wrote:

I’ve tried to put together a document describing a proposal for migrating uname sensibly (as well as providing the justification for doing so) with every attempt to minimize compatibility problems.

http://wiki.illumos.org/display/illumos/Modernizing+Uname

At least one user has indicated he’s afraid of the email storm. So am I, to be honest. So unless you have something really constructive to add — and have READ the document and UNDERSTOOD the ENTIRE proposal, please refrain from comments unless they will actually contribute new thoughts or ideas. Let’s not have a giant time wasting thread, ok? Email me directly if you’re not sure.

That said, I do want thoughtful and constructive critcism of the proposal. Is it unworkable? If so, why? Do you have a better idea? (Also, let’s keep this to people who will be *impacted* by this — release engineers, software engineers, downstream distributors. I don’t want it to devolve into a marketing debate, please.)

I don’t think, as that one user indicated, that just putting our heads in the sand and continuing to remain SunOS 5.11 forever is the right answer. Our kernel is evolving, and we have no way to even talk about this evolution, and the confusion with Solaris is unfortunate. Its well past time we addressed this — in fact I’m sorry I didn’t make this proposal much earlier — probably this should have be done from day 1, instead of now. :-/

--
Garrett D'Amore
Sent with Airmail

illumos-developer | Archives | Modify Your Subscription