The logical state machine for general interrogation does not reflect TrgOps.GI states. Does this state machine assume that TrgOps.GI is true? Or does NO GI ALLOWED infer RptEna or TrgOps.GI is FALSE?
The text following the state machine tries to clarify but there are some conflicts with no stated precedence.
The following 3 statements for example give different results for the following truth table
TrgOps.GI RptEna GI #1 #2 #3
F F F Resp- Resp+ Resp+
F F T Resp- Resp+
T F F Resp- Resp+
#1 The initial state is “NO GI ALLOWED”. Within this state, any GI.request (for example GI set TRUE) shall cause a GI.response– to be sent.
#2 If a client attempts to set GI=FALSE, a positive response shall be issued. However, GI pending processing will not be impacted.
#3 If a BRCB contains TrgOps.GI = FALSE then an attempt to set the GI attribute to TRUE shall cause a SetBRCBValues positive response. No GI report shall be generated.
TPCL 1.8 BrN5 and RpN5 currently tests for:
TrgOps.GI RptEna GI Response Queue Rpt
F T T Resp+ No
Note that if TrgOps.GI=FALSE infers “NO GI ALLOWED” state, then BrN5 and RpN5 expecting a Resp+ conflicts with the GI.request(GI=TRUE) transition to GI.response-.
Proposal
A simple truth table would eliminate ambiguity. I don't know what the intent of the standard is but if statement #1 were to take precedence over #2 and #3, the the truth table could look like this:
TrgOps.GI RptEna GI Response Queue Rpt
* F * Resp- No
T T T Resp+ Yes
F T * Resp+ No
* T F Resp+ No
The transition from “NO GI ALLOWED” to “GI ALLOWED” would require TrgOps.GI=TRUE and RptEna=TRUE for this solution. The transition from “GI ALLOWED” to “NO GI ALLOWED” would require only RptEna=FALSE since the TrgOps may not be changed while the report is enabled.
Arguably
* F * Resp- No
Might need to be (as per “NO GI ALLOWED” GI.request transition)
* F F Resp+ No
* F T Resp- No
Since it seems inappropriate to fail a noop.
Discussion
Created
Status
14 Dec 15
In Force (green)
Accepted. State machine attached which will be in Ed2.1.