Re: Call for Community Feedback: Retiring IETF FTP Service

[Keith Moore <[e-mail]>]

On 11/13/20 12:43 PM, Phillip Hallam-Baker wrote:

> On Fri, Nov 13, 2020 at 12:01 PM Keith Moore 
> <[e-mail] <mailto:[e-mail]>> wrote:
>
>     I'm very opposed to this proposal.
>
>     FTP is a much better interface than HTTP for scripting, mirroring,
>     and remote file access (i.e. mounting an FTP server like a "share"
>     so that it can be accessed from one's computer just like any other
>     file system
>
> I disagree. The ability to mount HTTP file systems is actually built 
> into Windows. I don't think FTP is supported.

Windows?  Who uses that?  :)   People who actually want to get work done 
generally try to avoid it.

Get HTTP file systems supported by as many implementations and platforms 
as FTP use as a file system, and I'll assent to transitioning from FTP 
access to HTTP file system access.

>
> FTP is a very peculiar protocol, I have implemented it several times 
> and it was awful to do even before NATs got in the way. It is not 
> really a separate protocol, it is an extension of Telnet.

Without arguing about details, yes it's a peculiar protocol. But a 
subset that works with 99.99% of the computer systems in use today is 
not that bad, and there are several decades of accumulated understanding 
behind it.   None of which is terribly relevant to this discussion, 
since almost nobody actually has to implement FTP any more, certainly 
not the IETF.   All IETF has to do is put a server up, and plenty of 
good servers are available off-the-shelf.

> Oh and having to redo every transfer because the default mandated by 
> the spec was to damage the file assuming a charset conversion was 
> ridiculous even for the time.

Yeah, but nobody does that any more at least between reasonable 
systems.   And at least transfers between very dissimilar systems still 
work, even though they're not much needed any more.

> We are inventing the future here, not keeping the past alive.

A future of constantly assured obsolesence is not one I want to 
contribute to, and that's  exactly where the web is headed.  It's such a 
complex mess now that the number of independent effective client 
implementations is approaching 1.  If the web is the future, we're all 
doomed.    And throwing away things that work is incredibly shortsighted.

Keith