Internet Tools (InfoMagic)/Internet Tools.iso/dos_win/winsock/maillist/94-05.Z/94-05/[e-mail]_Tue Apr 29 15:00:44 1994.msg

Internet Message Format | 1994-05-31 | 6KB

Received: from bigblue.oit.unc.edu by SunSITE.Unc.EDU (5.65c+IDA/FvK-1.07) with SMTP
          id AA28425; Mon, 2 May 1994 10:02:15 -0400
Received: by bigblue.oit.unc.edu (AIX 3.2/UCB 5.64/4.03)
          id AA21256; Mon, 2 May 1994 09:41:14 -0400
Received: from GATEWAY by bigblue with netnews
    for [e-mail] ([e-mail])
To: [e-mail]
Date: 29 Apr 1994 15:00:44 GMT
From: [e-mail] (Alun Jones)
Message-Id: <2pr7es$[e-mail]>
Organization: Welcom Software Technology
Sender: ses
References: <[e-mail]>, <2pm1pf$[e-mail]>, <2pna8u$[e-mail]>
Subject: Re: QVT 3.97 FTPd stops after bad login?

In article <2pna8u$[e-mail]> [e-mail] (Derek Chan) writes:
>In article <2pm1pf$[e-mail]> [e-mail] (Alun Jones) writes:
>>In article <[e-mail]> <[e-mail]> writes:
>>>I'm having a strange problem with the FTP server in QVT/Net 3.97... I
>>>have it setup serving anonymous servers, etc... and everything works
>>>fine, unless someone gets a failed login.  After a failed login, the
>>>server rejects *all* connection attempts.  It has to be quit, and then
>>>re-loaded.  Any ideas anyone?
>>
>>I'll bet a registered copy of my ftp server that this is because
>>you're using PC-NFS as your winsock stack.  They seem to have a major
>>problem with this.  This is at least one of the reasons why I don't
>>recommend use of PC-NFS unless absolutely necessary.
>
>Ummm....actually, I think Rick (Osterberg) and I are both using Trumpet Winsock beta- release 6...

>And it crashes.  What seems to be happening is that the ftp
>connection isn't closed properly and the WinQVT is limited to one
>connection at a time...
...
>P.S.  So....do I get that registered copy? :)

No - that offer was only made to Rick.  You can get an unregistered
copy, though, and with my blessing fill out the order form and send a
$15 cheque to register it! :-)

(For those of you who want it, this ftp daemon is at ftp.wst.com in
directory /pub/winsock/wftpd, or on many other popular winsock sites.)


Alun.
~~~~

--
The above is a personal opinion, and may not necessarily represent the
opinions of Welcom Software Technology, its management, or its staff.
For information about the WST lines of Project and Cost Management
Software, email [e-mail] or [e-mail], or call [phone]
From [e-mail] Mon May  2 10:28:26 1994
Received: from mpls.cnt.com (cnt.com) by SunSITE.Unc.EDU (5.65c+IDA/FvK-1.07) with SMTP
          id AA02334; Mon, 2 May 1994 10:28:26 -0400
Received: from cnt.com (hibbing.cnt.COM) by mpls.cnt.com (4.1/CNT-gateway-1.1)
    id AA11498; Mon, 2 May 94 09:15:37 CDT
Received: from smtpserv.cnt.com by cnt.com id AA14687
  (4.1/Ultra-1.4-10-10-91 for [e-mail]); Mon, 2 May 94 09:25:06 CDT
Message-Id: <[e-mail]>
From: [e-mail] (Esh, Andrew)
To: [e-mail] (Robert Reuss)
Subject: Re: Baud rates under Windows
Date: Mon, 02 May 94 09:24

>I have received a number of queries from folks who are using
>my external dialer, GoSlip!, regarding its maximum allowed
>baud rate of 19200.  The reason for this limitation is that I am
>using the serial port custom control included with Visual Basic
>3.0, and this custom control is limited to that baud rate.  Although
>I myself have a 14400 modem, I have never noticed any difference
>between using my program set to 19200 and other comm packages
>set to 57600, so long as I have TCPMAN set to 57600.  My
>assumption is that the modem negotiates the compression and
>acheives maximum throughput, but I could be wrong (it happens
>often)!  Moreover, under control panel I've noted that windows
>only offers up to 19200 as a port speed, but I'm told it will
>accept higher values????

>If anyone can educated me and others about the nitty gritty of
>baud rates under Windows, educate away!

>Regards to all!

>Rob Reuss
>[e-mail]


The reason baud rates are sometimes limited to 19200 bps on Windows
applications is that many PC's have serial port hardware that was
designed using an 8250 UART chip. This chip can only handle one character
at a time without being unloaded by the CPU. When Windows came out, it
was discovered that switching the CPU from one mode to another (Real
mode, Protected Mode), would keep the CPU busy long enough for the UART
to receive more than one character, causing it to lose all but the last
one.

Recently, more machines are built with modern UARTS such as the National
16550, which has a 16 character FIFO. It can accept characters until the
FIFO is full without any intervention from the CPU.

What all this means is that if you are running at greater than 9600 bps
on a machine that has 8250 UARTS (and many still do, even new ones), you
stand a good chance of dropping characters. With PPP, this will be
detected and the packet will be retried, but the failure rate could get
high enough to effectively tie the line up to a dead stop with
retransmissions.

My suggestions: Write software which allows baud rates as high as you
care to make them; those folks with good UARTs can use the higher rates.
If you want to get slick, check the UART type and limit data rate to
19200 bps if it's an 8250. There are a number of good Serial
Communications books which can show you how to detect the UART type.
"Serial Communications, a C++ Developer's Guide", by Mark Nelson, and
"C++ Communications Utilities" by Michael Holmes and Bob Flanders are two
good ones.

For users: Do not buy a PC which has an 8250 UART. You can find out the
type by running MSD (comes with Windows) and looking at the comm port
info. The bottom line shows the UART type. The National 16550 is probably
the best one to look for. If you already have a machine which has an
8250, you can buy a serial port card which has a 16550 to replace the
built in ports. Hayes has a really nice one out now, and other cards get
as cheap as $30.

Andrew
[e-mail]