Re: lynx-dev RFC959 non-compliance in Lynx hangs the client

[David Woolley]

> Note that it's a "should", and to me it sounds more like a general
> exhortation than a protocol requirement.

It's probably important to remember that this RFC was written before
the commercialisation of the internet and internet access software,
at a time when people implemented to the spirit, not the letter.

It is only in the later RFCs that it was necessary to use SHOULD, MUST,
SHOULD NOT and MAY to avoid commercial implementors implementing to the
minimum letter of the specification.

I would interpret this one as saying that half closed conditions should
be kept to a minimal duration.

> 
> [ 3.2  ESTABLISHING DATA CONNECTIONS ]
>       In general, it is the server's responsibility to maintain the data
>       connection--to initiate it and to close it. [...] The server
>       MUST close the data connection under the following conditions:
> 
>          1. The server has completed sending data in a transfer mode
>             that requires a close to indicate EOF.

I.E. the normal case.
> 
>          2. The server receives an ABORT command from the user.

To avoid the user's transfer thread stalling.

> 
>          3. The port specification is changed by a command from the
>             user.

Obvious.
> 
>          4. The control connection is closed legally or otherwise.

To make it clear that once the control connection is down there is
no FTP session.

> 
>          5. An irrecoverable error condition occurs.

Probably again to avoid hanging a blocked thread.