Stephen Hahn wrote:
> * Alan Burlison <[EMAIL PROTECTED]> [2007-03-06 15:16]:
>
>> Stephen Hahn wrote:
>>
>>
>>> I would agree that a project to rewrite kstat(1) in C requires
>>> investigation; in particular, an expected part of such a proposal
>>> would be a means of allowing access to raw kstats via the various APIs
>>> offered without simply having multiple copies of that information.
>>>
>> My personal preference would be for all the raw kstats to be replaced by
>> named ones, then the horrid "Platform x OS version x private header
>> file" mess which we now have would be gone. We'd have to keep the raw
>> kstats around for a while, but once they have gone consuming kernel
>> statistics in a platform-independent and release-independent way becomes
>> a *whole* lot easier.
>>
>
> It would be nice to get rid of raw kstats, but doing so breaks a bunch
> of software that knows about them. It would be nice to see a "this
> kstat supersedes that one" API. Otherwise, we just end up with both a
> raw and a named kstat for each bundle of information. Or we could
> just break some stuff. Maybe it's time.
>
I heartily agree with this.
How many tools depend on the binary kstats, I wonder? The ones in ON
can be fixed easily.
The other ones are harder, but I'm not sure any promises were made about
kstats ... i.e. I don't think they represent any kind of committed
interface.
The main tool vendors (e.g. CA, etc.) can probably adapt pretty easily.
It would be easy to provide a release where both stats were provided,
but that we announced EOF of the older RAW stats, as well.
-- Garrett
_______________________________________________
opensolaris-code mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/opensolaris-code
