Mr cURL wrote:
>
>On Sat, 21 May 2005, Georg Wicherski wrote:
>
>> it would be great if you could introduce an option in libcurl3 to disable
>> the sending of the TYPE command in passive connections or at least for
>> ignoring the response to it. This request is related to
>> http://www.mwcollect.org/ticket/30 .
>
>I've said it before, and I'll said it again: I would welcome a way to
>explicitly disable the sending of individual FTP commands. I've suggested an
>approach similar to how (default) HTTP headers are disabled.
>
>I've not had much response nor have anyone offered to do the work.
>

In mid-June I plan on playing with ftp to work on a VMS specific issue.
Since I'll be knee deep in the code at that time, I'll keep this suggestion
under consideration.

>> The two most widespread ftpd's around (sasserftpd and (u)r(x)bot ftpd)
>
>Are they really? curl has been out for more than 7 years and it has always
>sent this command and yet you are the first who brings this up!
>
>> send bogus replies (`226 Ok') to the TYPE command witch makes libcurl
>> disconnect.
>
>We've seen similar issues with other commands before. I think the better fix
>is to allow any 2xx response as a positive response to TYPE. Simply skipping
>TYPE seems wrong... Don't you agree?

In my opinion positive responses should not cause a disconnect. A log
message maybe, but not a disconnect.

Marty
Received on 2005-05-24