One way to surface potential users is perhaps to run an experiment and turn off FTP for 1 week (or whatever period of time) or to repoint the FTP service for that interval to a directory with a single text doc saying we're considering turning off FTP and please email us and tell us about your usage requirements. But this suggestion aside, we do appear to have quite clear usage data to support decision-making.
JL
On 11/18/20, 9:37 AM, "ietf on behalf of Joel M. Halpern" <ietf-bounces@xxxxxxxx on behalf of jmh@xxxxxxxxxxxxxxx> wrote:
There seem to be two groups of users causing the discussion.
One set are folks who use scripts that will be discommoded if we drop
FTP access. That is a concern. But a somewhat manageable one. And one
where we havve to at some point be able to say "no, we do not supporting
things forever".
The other argument is that there exist a set of people who will be
unable to practically get the documents if we drop the FTP access. If
true, that is important. But we do not appear to ahve any way to
evaluate the statement as other than a hypothetical. We know such
people could exist. But do we know if they do exist?
It seems to me that arguing for keeping the service because people in
the second category may exist is a very weak argument.
Yours,
Joel
On 11/18/2020 9:26 AM, Keith Moore wrote:
> On 11/18/20 9:21 AM, Lou Berger wrote:
>
>> How about: the FTP Service will be discontinued when FTP (RFC959) is
>> moved to Historic status.
>>
> One of the problems I have in general with reclassifying things as
> Historic is that this can lead to deprecation of services that are still
> quite useful, without the implications of such deprecation even being
> discussed, and without providing any advice about how to transition.
>
> "Historic" may be the wrong way of describing the situation. More
> accurately, a protocol may change from mainstream use - where it's the
> widely accepted way of doing some particular thing - to occasional or
> fringe use, where it's still useful in corner cases even though it's no
> longer the widely accepted way of doing that thing.
>
> Keith
>
>
>
