Le 29 déc. 2014 16:59, <mmann78@xxxxxxxxxxxx> a écrit :
>
> I was looking to add an FT_ enumeration for Fibre Channel addresses. See https://code.wireshark.org/review/6098/ for my attempt. Because the Fibre Channel address already had an "address type" (AT_FC), I thought a corresponding FT_ was appropriate. That seems to be how many of the "address types" are turned into field types. The review comments so far suggest that maybe an FT_ enumeration isn't the way to go, so I thought I'd pose the question to -dev.
>
> A Fibre Channel address is a 3 byte value, displayed with a decimal between each byte, displayed as hex values (ie ff.ff.ff). It does not have a "name resolution" component (like IP or Ethernet addresses).
>
> So if you're creating an hf_ item for it, I believe any of the following could be the way to represent it.
>
> 1. FT_FC, BASE_NONE (current approach)
> 2a. FT_UINT24, BASE_DOT (Suggestion that BASE_DOT would but a decimal between each byte value). proto_tree_add_item using the hf_ field would need a ENC_BIG_ENDIAN parameter.
> 2b. FT_UINT24, BASE_HEX|BASE_DOT (to ensure bytes are represented as hexadecimal. So an IPv4 address could be considered FT_UINT32, BASE_DEC|BASE_DOT if not for the name resolution)
> 3. FT_BYTES, BASE_DOT (perhaps other address types could just be different "punctuation" BASEs between their byte values)
>
>
> I'm looking for the "best", or at least "most consistent" approach. I also don't mind taking the time to change other existing methods to be able to identify/keep that consistency. Big picture is trying to cleanup address_to_str functionality as some of the comments in the code suggest.
>
Hi,
On my side I like option 3.
Regards,
Pascal.
