Re: [ftpext] FWD: ftp/959 reboot

[Paul Ford-Hutchinson <[e-mail]>]

Anthony Bryan wrote on 15/08/2010 20:50:27:

> 
> current implementations are probably mostly fine. the way things are,
> someone working on a new FTP app wouldn't even know 1123 existed.
> 

My experience of creating an FTP server from scratch in 1994 was that:

RFC959 was obviously useful
RFC1123 was just a list of things that I had to implement even if it made 
no sense in my environment (this is a _good_ thing from a standards and 
interoperability point of view)
There were some other RFCs that were also useful, I don't remember exactly 
how though.

To actually get it to work, I had to do an awful lot of dumping of tcp 
traffic between existing clients and servers because the spec (RFC959) is 
nowhere near defined enough to write an implementation.  (for example, the 
relative timing of the opening/closing of the data connection and the 
related FTP control connection commands is critical but undocumented, 
which is why I put them into RFC4217.)

I guess that the point I'm trying to make is:
- to write an implementation, you need to do more than just read the 
obvious RFCs, whilst we can and should help people to find the required 
information, in order for them to get to end of job coding such a beast 
will require either a huge documentation effort on our behalf, or a bit of 
gumption on theirs.

Paul