pcp
[Top] [All Lists]

Re: [pcp] Proposal for handling dynamic metric names (and hence dynamic

To: Martin Hicks <mort@xxxxxxxx>
Subject: Re: [pcp] Proposal for handling dynamic metric names (and hence dynamic metrics)
From: Mark Goodwin <goodwinos@xxxxxxxxx>
Date: Fri, 10 Jul 2009 12:23:25 +1000
Cc: Nathan Scott <nscott@xxxxxxxxxx>, kenj <kenj@xxxxxxxxxxxxxxxx>, pcp@xxxxxxxxxxx
In-reply-to: <20090709122558.GB5068@xxxxxxxxxxxxxxxx>
References: <4A5541FE.9090905@xxxxxxxxx> <144401009.303101247103013280.JavaMail.root@xxxxxxxxxxxxxxxxxx> <20090709122558.GB5068@xxxxxxxxxxxxxxxx>
User-agent: Thunderbird 2.0.0.21 (X11/20090320)
Martin Hicks wrote:
On Thu, Jul 09, 2009 at 11:30:13AM +1000, Nathan Scott wrote:
And having a 'pad' field in the middle of a structure seems kind
of funky, but I can see the reasoning for wanting it there. Maybe
just use up the two existing pad bits and call it 'flags'? (with
room for three more flag values in the future, one of which could
be to flag an extended range of domain values).
Hmm, I'd be more inclined to extend "domain" now - we've used up
more than half the available domain numbers already, so this seems
like the most useful way to use that bit IMO (256 numbers looks a
tad small to me, nowadays, whereas 512 would buy alot of headroom).

Yeah, I agree with this.  We might as well take this step now to leave
plenty of room for growth.

With the int flags:2 approach, we'd still have one unused flag combo
available for future use. With the domain extension, we'll have zilch.
Probably should've made the darn things uint64 in the first place!

Just a thought :) I'll go with the domain extension.

Cheers

<Prev in Thread] Current Thread [Next in Thread>