Hi,

a short comeback to this list for a few infos. Sorry if I don't stay too long, I 
usually
don't have the time to deal with the amount of traffic here.

Quite often I receive questions about the mentioned subjects. So maybe it's good to
clarify a few things for a wider audience.

QLwIP, what works:

 * HTTP server, including support for PUT (upload) method
 * Graphical Mail client: POP3, SMTP with authentication, local folders,
   serverside message handling, multiple accounts
 * Simple TCP telnet/echo test server
 * Hardware driver for Q40/Q60 ethernet
 * Hardware driver for serial (SLIP)
 * Ping echo
 * Supported protocols: ICMP, IP, UDP, TCP
 * C API similar to Berkeley socket API, _only_ internal use
 * Multitasking use of the stack by several jobs

QLwIP, current restrictions:

 * Applications and library must be linked to the same binary
 * Final API not specified nor implemented
 * Jobs using QLwIP can only be savely removed if the network is brought down
 * Name resolver available, but not intergrated
 * TCP over ethernet is artificially limited to about 75 kBytes/sec throughput

QLwIP Release:

The software can already do useful things like transferring files by TCP
protocol, which is a good thing especially when communating with a
Windows or Linux box, or act as a Webserver, or deal with personal
email if one has a DSL or ISDN router. E.g. this mail was sent by QLwIP
from a Q60 under QDOS Classic (Have a look at the "X-Mailer" header ;-) 
Still I don't find QLwIP fit for a release. Why?

 * QLwIP and especially the API have not the final structure. A premature
   release could mislead other developers who want to write TCP/IP apps.
 * The OS in it's current stage doesn't allow fullspeed multitasking TCP if
   the ethernet driver is interrupt triggered. An OS change is required, before
   I can give QLwIP the preferred structure.
 * I would like to see QLwIP _integrated_ into an operating system and use a
   free software license for my work. Now SMSQ/E is incompatible to all free
   software licenses, Minerva is free software but not ported to Q60,
   and QDOS Classic has only a small user base. This is kind of a dilemma.

USB:

Same old question. It would be feasible to develop an USB hardware add-on
for Q40/Q60. USB host controllers with simple bus interfaces are available,
a card that fits into a Q60 extension slot or maybe ROM socket is not a very
hard design challenge. Nevertheless I won't make such a hardware, because it
is absolutely unrealistic that the necessary QL software will be written.

Q60 Successor:

I admit, it's true that D&D Systems and me had internal plans for a Q60
successor. They included a new Coldfire derivative with FPU, and a new
approach to overcome remaining incompatibilities between Coldfire and 68k.
I gave up these plans, because they would have required SMSQ/E to be
modified and optimized for the new hardware. After the departure of Tony Tebby,
I see no basis for projects like this anymore, because there's no common ground
with the new SMSQ/E maintainers, and at the same time no freedom to develop
SMSQ/E under open source conditions. QDOS Classic provides this freedom, but I
currently find it not advanced enough to justify another large investment of my
time and money in QL hardware. So from my side there's no hot product in the
pipeline.

All the best
Peter


  • ... Peter Graf
    • ... gwicks
      • ... Peter Graf
    • ... wlenerz
      • ... Fabrizio Diversi
        • ... Jerome Grimbert
          • ... "Phoebus R. Dokos (Φοίβος Ρ. Ντόκος)"
            • ... John Taylor
              • ... "Phoebus R. Dokos (Φοίβος Ρ. Ντόκος)"
              • ... wlenerz
            • ... Marcel Kilgus

Reply via email to