pcp
[Top] [All Lists]

RE: [pcp] Duplicate Names in the PMNS - seeking consensus

To: "'Nathan Scott'" <nathans@xxxxxxxxxx>
Subject: RE: [pcp] Duplicate Names in the PMNS - seeking consensus
From: "Ken McDonell" <kenj@xxxxxxxxxxxxxxxx>
Date: Wed, 21 Jan 2015 10:48:57 +1100
Cc: "'PCP'" <pcp@xxxxxxxxxxx>
Delivered-to: pcp@xxxxxxxxxxx
In-reply-to: <1411235146.14188979.1421795599254.JavaMail.zimbra@xxxxxxxxxx>
References: <54BA00E9.7050504@xxxxxxxxxxxxxxxx> <1411235146.14188979.1421795599254.JavaMail.zimbra@xxxxxxxxxx>
Thread-index: AQDP3cuxAcpXF3/4aO/vlFsSs1/IUgFUEXvinr+3q5A=

> -----Original Message-----
> From: Nathan Scott [mailto:nathans@xxxxxxxxxx]
> ...
> Could you send a link to the "ugly hack" branch?  It'd be good to know just
> how invasive the needed changes are before we make a call.

The "hack" is for option 2 that I don't favour ... it is a private branch in a 
local repo.  I could send you diffs, but it would not be pretty.

If we go for option 1, the changes are much, much smaller because 
pmLoadNameSpace would call pmLoadASCIINameSpace with the "allow dups" option, 
and so I would not have to change pmLoadNameSpace to pmLoadASCIINameSpace 
everywhere ... 8^)>

> The only namespace trickery that I have seen used "in the wild" is the
> dynamic namespace agent re-routing thing, as implemented in pmdasample
> (sample.secret.foo.bar.max.redirect) already.  That's been used in one
> production deployment I know of.

I don't think this would be impacted at all ... this is disjoint from the PMNS 
loading functions and their users which where the pain is located.

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