RE: run.stableId

From
Larry Golding \(Comcast\) <>
Date
2018-04-24T20:25:54+00:00
ID
[email protected]
Thread
RE: run.stableId
Sounds good.

 

I think we should allow multiple levels, so either of these is ok – it’s up to the engineering system to decide the appropriate level of bucketing:

 

Nightly security run/x86 debug

Nightly security run/x86/debug

 

Larry

 

From: Michael Fanning <> 
Sent: Tuesday, April 24, 2018 12:51 PM
To: Larry Golding (Comcast) <>; 'James A. Kupsch' <>; 
Subject: RE: run.stableId

 

I think we should add it, for precisely the reason you cite, bucketing at the stable id level. You might do this in order to organize all tool breaks for both the x86 and amd64 instances of a tool run.

 

Nightly security tools run for Contoso service

X86 : 0 breaks 

Amd64: 10 active issues

 

From: Larry Golding (Comcast) <> 
Sent: Friday, April 20, 2018 11:17 AM
To: Michael Fanning <>; 'James A. Kupsch' <>; 
Subject: RE: run.stableId

 

Ping on this. Do we really need the namespacing, given my argument below?

 

From: Larry Golding (Comcast) <> 
Sent: Wednesday, April 18, 2018 12:19 PM
To: Michael Fanning <>; 'James A. Kupsch' <>; '' <>
Subject: run.stableId

 

As I understand it, the goal is to bucket together runs of the same type, so that a result management system can:

 

Open a SARIF log file,

Find run.stableId,

Find the most recent run of the same type, and

Copy run.id from the earlier run into run.baselineId of the current run.

 

That doesn’t require namespaced tags. It works whether run.stableId contains "Nightly security tools run/x64 release optimized" or whether it contains "Nightly security tools run x64 release optimized".

 

The only reason you’d need namespacing is if there’s some other scenario that needs to bucket together every "Nightly security tools run", regardless of the build configuration.

 

Larry