On Mon, 18 Jul 2011, Terry Manderson wrote:
On 18/07/11 9:39 PM, "Tim Bruijnzeels" <[email protected]> wrote:
Hi,
I agree that not having this mapping is tedious and error prone for RPs.
I can agree that a mapping system is useful. It may just be that living unix
world for far too long has seen me move away from the mandatory dos-like
suffixes to the voluntary use of extensions in a unix file system as a
*hint* to the file contents and nothing more.
And I'm happy to see it written as a hint. A validated mapping should come,
in my opinion from something more robust which also transcends the
technology used in the repository.
There was a brief discussion of the use of file names extensions when the
repos-struct document came up for last call. See the following messages:
http://www.ietf.org/mail-archive/web/sidr/current/msg02281.html
http://www.ietf.org/mail-archive/web/sidr/current/msg02282.html
http://www.ietf.org/mail-archive/web/sidr/current/msg02283.html
http://www.ietf.org/mail-archive/web/sidr/current/msg02284.html
To summarize: George Michaelson spoke against extensions when we were
considering a registry (and Terry mildly supports them), I asked George if
he as author was suggesting the draft needed to change and he said no and
added that rsync can filter objects only on the basis of the file name.
So we've been around this barn already.
The point about the rsync filtering abilities has not come up this time.
I ask for review and consideration of that exchange (and the registry
discussion context).
--Sandy, speaking as wg chair
_______________________________________________
sidr mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/sidr
