Hi,

I recall having a discussion on this topic with Ernst Horvath when this clarification text was proposed. I thoroughly checked the ASN.1 back then and based on that I must add that both Sasha and Ernst are right.

Again, in the Invoke, the invokeId is encoded as a constrained integer. In the ReturnResult and ReturnError, the invokeId is encoded as "a regular" integer. I do not see any concerns for interoperability, unless it is your intention to copy the encoded invokeId from the Invoke straight to the ReturnResult/ReturnError. There is no inconsistency, between the different APDUs, there just happens to be a different specification for the invokeId in the different APDUs. This specification is identical to the one used in X.880, on which H.450.1 is based.

I don't quite understand what you mean with the "two ASN.1 versions" in the "does not gel in" section of your text. Coul you clarify this?

Regards,

Frank

Hi

I just want to rewrite the same to be sure we are in sync.

6.6.1.3 of the V4 implementors guide does not CHANGE/CONTRADICT ASN.1 of H.450.1 it EXPLAINS how to encode values correctly according to ASN.1 definitions of H.450.1 And this correct encoding is: (as written in 6.6.1.3) invoke.invokeId shall be encoded as INTEGER(0..65535) returnResult.invokeId and returnError.invokeId shall be encoded as INTEGER

Sasha

-----Original Message----- From: Horvath Ernst [mailto:ernst.horvath@SIEMENS.COM] Sent: Monday, May 13, 2002 8:49 AM To: ITU-SG16@echo.jf.INTEL.COM Subject: AW: H.450.1 ASN.1 interoperability issue

Mark, there was no intention to change any ASN.1 definition in H.450.1. The note was added just to make implementers aware that the InvokeID is encoded differently in the "invoke" APDU and in the "returnResult/returnError" APDUs: in the first case it is constrained by InvokeIdSet, in the second case it is not constrained. This is not an incosistency, just a consequence of the way Packed Encoding Rules were designed. Please also note that "InvokeIdSet" is used in two different meanings: in table 4 it is the name of a parameter of the parameterized structure ROS (i.e. a placeholder), whereas in table 3 it is an "actual" value-set of integers. The brackets in the new note (InvokeIdSet = 0..65535) are not an ASN.1 definition, just an indication what the effective constraint is, namely 0..65535. Table 3 is still the (unchanged) formal definition of the actual "InvokeIdSet". Ernst Horvath Siemens AG

-----Ursprüngliche Nachricht----- Von: Mark Brown [mailto:Mark.Brown@ED.ACULAB.COM] Gesendet am: Freitag, 10. Mai 2002 17:29 An: ITU-SG16@echo.jf.INTEL.COM Betreff: H.450.1 ASN.1 interoperability issue

We've been experiencing some fairly serious interoperability problems with some systems due to confusion over what exactly the ASN.1 for H.450.1 is and were hoping someone could clarify things for us. The problem stems from section 6.6.1.3 of the V4 implementors guide which adds a note below table 4:

In the Invoke APDU, the invokeID is an INTEGER constrained by a PER-visible constraint (InvokeIdSet = 0..65535) and is therefore encoded as a *constrained* INTEGER (16 bits, no length field). In the ReturnResult and ReturnError APDUs, however, the invokeID is encoded as an *unconstrained* integer (with explicit length field) because the applicable constraint... is not PER-visible.

However, in H.450.1 table 3 (on page 8) defines InvokeIdSet to be:

InvokeIdSet INTEGER ::= {InvokeIds,...} InvokeIDs ::= INTEGER (0..65535)

which does not gel with the text in the IG and causes interoperability issues since the two versions of the ASN.1 for InvokeIdSet are not encoded identically. This text appears to be unchanged from the prepublished edition on Packetizer.

Our understanding of the text added in the IG is that the intent was to emphasise that the encoding of the invokeID is not consistent between the various APDUs and that the no change to the ASN.1 was intended. This is particularly the case since the IG doesn't modify table 3 as would be expected if it were correcting the ASN.1 there. Others read things differently and are using the definition of InvokeIdSet given in the IG.

What is the intention here? If the intention is to redefine the ASN.1 then some clarification as to why this was done would be very much appreciated.