From: <sk...@ki...> - 2007-05-21 21:03:09
|
On Mon, 21 May 2007 16:47:40 -0400 (EDT) "Stuart D. Gathman" <st...@bm...> wrote: >Pyspf currently expects to get str for IP6 addresses. I *think* the >AAAA support added (not by me) will break this by returning a human >readable form instead. I guess they wanted to be consistent with >A and getaddr(), which returns a human readable form. > >1) What is the purpose of 4*I format for IP6 addresses? If you're >masking and stuff (which pydns isn't), then long is what you want (and >that is consistent with addr2bin for IP4). If you want compact, then >a simple binary str is what you want. Even 8*H would make pure python >conversion to human readable form simpler. But 4*I ? Should we change pydns >to use str as the binary form internally? > >2) Should I change pydns to return binary str for AAAA to be consistent >with 2.3.0? Or change the pydns driver in pyspf to check version? >Perhaps an option to request binary str data (maybe it's already there)? >Unfortunately, some 16 byte IP6 addresses look like a human readable >form as ASCII, and vice versa - so auto-detection is out. > It's not just pyspf. I'm the Debian maintainer for pydns and an uploader (asst. maintainer) for pyspf and in Debian pyspf accounts for only a small part of pydns usage. I feel VERY strongly that you should not break an interface that has been stable for years. If you need to add new stuff, add it without breaking the old. BTW, I've done a number of patches I'll forward to the list. Scott K |