ofl@kos.rci.dk (Ole Frank Larsen) (05/28/90)
There seem to be some misunderstandings of both my original questions, and the interpretation of the ISO8348/Add.2. In article <30641@cos.com> cos.UUCP!grieve (David Grieve) writes: > >What is the difference between an NSAP address with AFI = 36 and > >and NSAP address with AFI = 37? > > The encoding is the difference. It is too long to go into so > read carefully clause 8.3, then use the binary encoding anyway. And UVCW.UVIC.CA!eskovgaa (Erik Skovgaard) writes > I beg to differ. All X.25 addresses are normally encoded in BCD; i.e. > two decimal digits per byte. > > The differences between the AFIs 36 and 37 are in the encoding of the > DSP which may be used in a subnetwork off, say a public X.25 network. > Such a subnet could be either X.25-based - which would mean that you > should use AFI=37 - or it could be a LAN which mean that you should be > using AFI=36 (assuming the LAN uses binary encoded addresses). This is wrong. The AFI does not say *anything* about the coding of the DSP. The AFI only specifies the abstract syntax of the DSP, NOT THE ENCODING! The encoding of the DSP is left to the protocol used, and X.25 specifies the the preferred binary encoding of 8348/Add.2 should be used. Binary coding of decimal syntax, that ends up being BCD, which is what is used for the IDP of *all* NSAP-addresses, whenever binary encoding is used. The choice of encoding does not prevent the DSP being specified in decimal syntax, as long as it is *encoded* binary. This is what puzzled me in the first place. If the AFI does not say anything about the encoding of the DSP, then why have two AFIs that only specify different *syntaxes*? The second issue here is regarding routing information in the NSAP-address. ISO 8348/Add.2 states this very clear: "Network service users cannot [read: ought not to] derive any routing information from an NSAP address." Thus, I am perfectly aware that an NSAP-address should be treated as an octet string. All routing should be done according to the ES-IS and IS-IS protocols. BUT, why then have the AFI specifying the abstract *syntax* of the DSP?? PS. Erik skovgaard suggested that there was 4 different AFI values allocated for the X.121 address-domain. This is true. AFI values 52 and 53 corresponds to AFI 36 and 37, the only difference being whether leading zeroes in the IDI should have any significance. -- Ole Frank Larsen. RC International, Denmark ofl@rci.dk Hackers do it with fewer instructions