Re: lynx-dev ftp dir listing

[Thomas Dickey]

On Tue, 6 Jan 2004, Gisle Vanem wrote:

> > This doesn't appear to be the same case as the broken ProFTPD.  I'm
> > not sure how to analyze it.
>
> I did this little patch which seems to help:
>
> --- htftp.c.org   Tue Jan  5 22:27:40 2004
> +++ htftp.c     Tue Jan  6 11:13:44 2004
> @@ -3195,10 +3195,15 @@
>                 *cp2 = '\0';
>             }
>             status = send_cmd_1("PWD");
>             if (status == 2 && response_text[5] == '/') {
>                 status = send_cwd(filename+1);
>                 if (status == 2) {
>                      StrAllocCopy(fn, (filename+1));
> +                    type = "D";
> +                    isDirectory = YES;
>                      if (cp2) {
>                         *cp2 = '/';
>                         if (fn[strlen(fn)-1] != '/') {
>
> Not sure what the intention is behind that ugly piece of code, but my guess
> it tries to detect a CWD != "/~user". So if it expands it must be a dir, 
> right?

right (I don't see enough context to know if the change is correct, but
will study it).

> It also worked prior the patch if I did 
> ftp://giva:address@hidden/~giva/;type=D
> but that should be needed IMHO.
>
> But there's another (unrelated) problem. The comment at top of
> htftp.c says "A cache of control connections is kept", but that's
> not what I'm seeing. Lynx closes and reconnects for every directory
> visited. From ftp://ftp.deepspace6.net/:

I think the comment (and some of the logic) is not currently accurate.
The code is maintaining some data which is thrown away - would be
interesting to know more of the history of that.

>
> HTFTP.c: 2916: HTFTP: Closing data socket 16
> HTFTP.c: 471:     Rx: 226-Options: -l
> HTFTP.c: 471:     Rx: 226 3 matches total
> HTFTP.c: 3793: HTFTPLoad: normal end
> HTFTP.c: 3797: closing control socket 18
>
> Better seen in Ethereal:
>  deepspace6.net -> 10.0.0.7     FTP Response: 226-Options: -l
>     10.0.0.7 -> deepspace6.net TCP 1391 > ftp [FIN, ACK] Seq=3644834859 
> Ack=2734810633
>  deepspace6.net -> 10.0.0.7     FTP Response: 226 Logout - CPU time spent: 
> 0.000 seconds.
>
> Why? Is the "226.." confusing Lynx? Don't think so because the code at
> bottom of HTFTPLoad() does the NETCLOSE() every time AFAICS.
>
> BTW. I used FTP_PASSIVE:FALSE since that's seems to be a lot faster
> with my NAT and most ftp-servers.

I expect that's because it doesn't have to make the socket work in full
duplex.

-- 
Thomas E. Dickey
http://invisible-island.net
ftp://invisible-island.net

; To UNSUBSCRIBE: Send "unsubscribe lynx-dev" to address@hidden