Pete, Bancroft, Pekka and others,

now as I'm back from my long vacation I want to throw in my bits to this discussion. (By the way, I will try to answer or comment all the other pending emails which have been sent to me).

1.) RandomVal/INTEGER: It is true that H.235 has defined this as an arbitrary long integer for some reason which I can't remember. I've heard also that there are implementations that can only process 32-bit integers. H.323 Annex J uses RandomVal only as an increasing sequence number/counter. Thus, Annex J does not really require RandomVal as an arbitrary long integer; a 32-bit integer would be sufficient for that purpose while still guaranteeing sufficient security. H.235 Version 2 could restrict the scope of the RandomVal Integer to 32 bits; if this is considered useful and we can all agree to this. Please let me hear your opinions on this.

2.) nonStandardParameter: I can only guess why H.235 has chosen a simpler definition than H.225.0: Apparently, H.235 does not need the H.221 non-standard support; this was considered not necessary. Thus h221NonStandard was left out in the structure leaving the Object identifier. I hope that this simplification does not cause real implementation difficulties.

3.) I do not quite understand why you consider object identifiers limited? In my opinion, the concept of object identifiers is quite flexible and targeted for interoperability. H.323 Annex J section 8 shows a proposal for the various object identifiers used.

Regards,

Martin. ----------------------------------------------------------------------- | Dipl.-Inf. Phone: +49 89 636-46201 | Martin Euchner Fax : +49 89 636-48000 | Siemens AG | ZT IK 3 mailto:Martin.Euchner@mchp.siemens.de mailto:Martin.Euchner@mchp.siemens.de | Intranet: http://zt-security.mchp.siemens.de/Standardization/ITU-T_SG16/index.html http://zt-security.mchp.siemens.de/Standardization/ITU-T_SG16/index.html | Otto-Hahn-Ring 6 Internet: http://www.siemens.de http://www.siemens.de | D-81730 Muenchen | __________________ | Germany -----------------------------------------------------------------------

-----Original Message----- From: Pete Cordell [SMTP:pete@TECH-KNOW-WARE.COM] mailto:[SMTP:pete@TECH-KNOW-WARE.COM] Sent: Wednesday, September 08, 1999 9:26 AM To: ITU-SG16@mailbag.cps.intel.com mailto:ITU-SG16@mailbag.cps.intel.com Subject: Issues with H.235

Dear All,

I'm implementing some of the H.235 stuff and have a few concerns.

RandomVal is defined as INTEGER only. This is not a particularly helpful definition as in theory this could be a million bit + integer if needed. Not many computers support such types! In fact, a well known ASN.1 compiler maps this to an int which is a signed 32-bit value on our platform. Is this sufficient? Without further discussion about the range of this value I feel there is a potential for interoperability problems.

Perhaps we can say that RandomVal will never be more than 32 bits long, and then add a type like RandomSeq as an OCTET STRING for cases when we need a longer random value.

There are also a few other issues, for example:

nonStandardParameter in H.235 is defined differently to that in H.225. Why is that?

Similarly tokenID only takes an OID. Again, why such a limited format?

Regards,

Pete

============================================= Pete Cordell pete@tech-know-ware.com mailto:pete@tech-know-ware.com =============================================