I completly disagree.
Just today I installed FreeDOS on a brand new Asus board with SATA2 a
gigabit ethernet chip.
Simple: go to www.netbootdisk.com and create a floppy. After it boots
and detects the NIC, copy it's driver including packet driver. Easy...
The truth is that if you have an application that is worth using with
FreeDOS, drivers exist :)
And it is damn fast
Em 06-08-2010 22:22, Someone escreveu:
> I'd say that Freedos is okay for networking if you have a card that is
> supported by open source drivers, but that seems to be a very short list
> of cards at this time. I guess the question to ask of anyone seeking to
> network Freedos is, is a dos environment enough for what you are after?
> Command line linux is much easier to network than freedos and there is a
> lot of utility to it. I don't foresee a strong effort to get free
> network driver support to track with today's gigabit and 10 gigabit
> built-in and PCI cards. Companies aren't releasing dos drivers for the
> most part anymore because Microsoft isn't supporting DOS. This is a
> shame, Microsoft shouldn't have to support an OS for that OS to be
> I question whether TCP/IP is the best way to go in a DOS environment.
> Local area only networking with special filtering gateways to deal with
> remote data and remote access makes more sense security wise than
> hooking DOS directly to the Net via TCP/IP. What is your application?
> Are the DOS based clients seen as single user terminals hooked to a
> central network server that handles authentication, routing, providing
> network services, etcetera? Are DOS based clients going to access each
> other for services?
> Novell's Netware, up until the company gave up and went to a TCP/IP
> based system, was a local area network system. There are many
> disadvantages to running the same protocol that the Internet runs for
> local area networking. First problem that strikes immediately is
> addressing. Where IPV6 is supposed to help, most people don't have
> IPv6 equipment where IPv6 is unheard of in DOS environments. Even
> so, IPv6 is just a band-aid measure if globally unique IP addresses
> continue to be given out thoughtlessly.
> Services like DNS, dhcp, etcetera don't seem to be designed to
> distinguish between a local verses a remote source. Routing is
> overly complicated if a private network is TCP/IP based. Say I
> assume that the .home or the .pri domain are not used on the Net.
> Maybe I think that .cox isn't used. There is no way to set DNS
> up to distinguish a local use of .foo when there is the
> possibility of a remote usage of foo. Where there is name
> collision, how do you sort that out? If you know you are looking
> for local resolution of a name and there is some way to indicate
> that in a standard manner, problem solved. Reality is, .foo which
> is not in use in Internet names today could be used tomorrow.
> Other problems come into play when you search by host name and
> failing to add the private domain name leads to an erroneous
> name resolution. There seems to be no standard for naming private
> TCP/IP based networks that are occasionally Net connected. Some people
> say use your global domain name on your private hosts. Other people
> say, buy two domain names and use one of them for the private hosts.
> Still others say, use a .cox, .pri, .home, or some .foo that is not
> in use on the Internet. All of these options have serious drawbacks.
> Using your global domain name on private hosts that may never touch
> the Net, you can't use those host names twice. Using a second
> globally registered domain for private hosts reduces the global
> name space. Trying to guess that domain foo will never be used
> on the Internet, there are no standard choices which means that
> the domain you choose could be adopted at any time. IPv6 doesn't
> help, most people can't implement IPv6 and even if you can, IPv6
> only increases the number of addresses without providing a way
> to let people reuse names/overlap names. The strings people come
> up with to name IP addresses can only be so long before it becomes
> easier to simply type in the number. I'm afraid that the naming
> convention in use 10 years from now will be a gross patch on top
> of what we do now.
> DOS environments are incapable of protecting the hardware as the kernel
> is NOT a layer around the hardware. QDOS stood for quick and dirty
> operating system and people should realize that DOS was never intended
> for a networked environment. DOS combined with an IPX/SPX network is
> interesting, but Netware is totally proprietary.
> Look at mailing addresses in the U.S. There are city level unique zip
> codes, at least 10^5 or 100k possibilities. That is assuming that all
> zip codes are maximum 5 digits long (which probably isn't true). Look
> at the street level though and chances are you can find an Elm street
> for example in 100's of places. The combination of street name, street
> address, and zip code will always be unique. Street names that are
> popular get reused, something that isn't easily done with the current
> Internet naming scheme. Does everyone want to type the equivalent of
> a complete mailing address to get to a web site? No, of course not.
> Maybe in the future, simple names will be expanded by tools that help
> you to expand them properly.
> A private network should route and name hosts independently of the
> Internet and getting to the Net, for security reasons, should
> involve a special procedure.
> My simple recommendation is, consider something other than TCP/IP
> for a DOS environment and Linux for Internet connected TCP/IP
> networks. For security reasons, most of us probably don't want
> our DOS environments to connect directly to the Net, but for
> gaming purposes, local area networking can be quite fun.
> This SF.net email is sponsored by
> Make an app they can't live without
> Enter the BlackBerry Developer Challenge
> Freedos-user mailing list
This SF.net email is sponsored by
Make an app they can't live without
Enter the BlackBerry Developer Challenge
Freedos-user mailing list