On Wed, May 27, 2015 at 09:02:55PM +0100, Paul Jakma wrote: > On Wed, 27 May 2015, David Lamparter wrote: > > I think we should have two fields: hw_addr_type, identifying the type of > > information that hw_addr contains, and hw/ll_type, identifying the link > > itself. > > Is there a need for 2 types? There's no way to get at lower-level hardware > or other addresses is there? E.g. a tunnel interface will present, say, an > IP address there I presume?
Best example: LLT_ETHER and LLT_IEEE80211 are distinct (and in fact we're working on some homenet foobar that might want to know if a link is wifi), but they both have ethernet addresses. (Then again, that homenet code has other external interfaces.) > > Rationale being, in rtadv/isisd, we don't care what the link type is, we > > just have code that checks for an (ethernet) address and uses that. > > Right. If it's ethernet presumably. Ethernet or wifi :) ... and the rtadv code could do something useful for EUI64 addresses too, which is quite a bunch in the list. (1394, IB, 802.15.4 - I think.) [list cut] > I'd probably keep X.25, (C)?SLIP(6)? and (RAW|C)HDLC. Still being used for > new stuff here and there I think. I don't think they are :) ... even if: is it really useful to have these values in Quagga, if we aren't doing anything with it either way? Some stuff - either hung old or flashy new - will be represented as UNKNOWN whatever we do... -David _______________________________________________ Quagga-dev mailing list [email protected] https://lists.quagga.net/mailman/listinfo/quagga-dev
