Server code for Win95 or NT
Eric Muehle
Does anyone know where one would find a server for Win 95 or NT?
Thanks,
Eric
Caesar
On Sun, 15 Mar 1998 19:03:48 -0800, "Eric Muehle" <emu...@msn.com>
wrote:
>Does anyone know where one would find a server for Win 95 or NT?
Doesn't exist...
Yet, although several people have played with the notion.
Daniel O'Connor
Caesar wrote:
> >Does anyone know where one would find a server for Win 95 or NT?
>
> Doesn't exist...
>
> Yet, although several people have played with the notion.
Yes, well, the code is truly ugly.. It in fact would be quite
difficult to compile it under Win 95/NT.. Maybe using Cygnus tools you
could do it..
Hey, perhaps someone could rewrite it? :)
Darius
Alec Habig
Daniel O'Connor <doco...@nospam.gsoft.com.au> wrote:
>
> Yes, well, the code is truly ugly.. It in fact would be quite
>difficult to compile it under Win 95/NT.. Maybe using Cygnus tools you
>could do it..
Forget 95, you really need a multitasking operating system to do a
server. Context switches are the big server bottleneck.
For NT, ugliness aside, there are some fundamental problems with the way
NT handles process communications. The server has shared memory at its
heart, NT doesn't implement this in a compatable way to the rest of the
world. A second but more solvable problem is NT's lack of a good vfork().
>Hey, perhaps someone could rewrite it? :)
It would be a ground-up rewrite, changing the fundamental structure of
the code, and it might not work on other systems when you're done, but
sure, you could in principle do it.
--
Alec Habig, Boston University Particle Astrophysics Group
ha...@budoe.bu.edu
http://hep.bu.edu/~habig/
Alec Habig
Daniel O'Connor <doco...@nospam.gsoft.com.au> wrote:
> Yes, well, the code is truly ugly.. It in fact would be quite
>difficult to compile it under Win 95/NT.. Maybe using Cygnus tools you
>could do it..
As an aside an possibly unwarrented flame bait, I have a theory about
why development for Windows and MacOS have lagged behind in netrek
programming.
Programming clients and servers takes both a fair bit of know-how and a
desire to tinker with source code (as well as spare time to burn).
People who know what they're doing and have an itch to mess with the way
programs work aren't using Windows for both reasons.
o It's plain not as good as other OS's, it's only redeeming value is
its sheer universalness. Everyone knows it, writes commercial apps
for it, and any dummy can do something useful with it. Thus, most
people with know-how gravitate away from it, as they know how to get
better use of the alternatives - unless they're writing those
universal apps for money (in which case, they don't have the time to
hack for free).
o It's a stifling environment - you must do things Microsoft's way.
Thus, it's not possible to fiddle with the way it works to the same
degree as other systems, particularly the open ones. This drives
away the people with the urge to tinker.
So, we're left with few people in the union of the sets
{people who want to and are able to work on netrek code}
and
{people who will do it on Windows}
and when those few people {the set of Shekter and Collenberg} lose
interest or get busy, nothing happens on the Netrek for Windows front.
The message to all you people whining about it is this - download the
source and put some work into it. Prove me wrong.
Tom Holub
In article <6eorgk$1e0$1...@news1.bu.edu>, Alec Habig <ha...@budoe.bu.edu> wrote:
)
)The message to all you people whining about it is this - download the
)source and put some work into it. Prove me wrong.
Or install linux.
-Tom
Ra\/en
Tom Holub wrote in message <6ep6n4$pkf$1...@shell3.ba.best.com>...
Is linux faster to run? I'm guessing it would be, but I'm not sure.
Ra\/en
Alec Habig wrote in message <6eoq2m$lt$1...@news1.bu.edu>...
>Daniel O'Connor <doco...@nospam.gsoft.com.au> wrote:
>>
>> Yes, well, the code is truly ugly.. It in fact would be quite
>>difficult to compile it under Win 95/NT.. Maybe using Cygnus tools you
>>could do it..
>
>server. Context switches are the big server bottleneck.
>
>For NT, ugliness aside, there are some fundamental problems with the way
>NT handles process communications. The server has shared memory at its
>heart, NT doesn't implement this in a compatable way to the rest of the
>world. A second but more solvable problem is NT's lack of a good vfork().
Why don't we just ask our good friends in Redmond to impliment that stuff in
NT 5.0? I'm sure they would do that if it would help them rake in more $.
Tom Holub
In article <6epa63$8ou$1...@usenet85.supernews.com>,
Ra\\/en <ra...@SPAMTHIS.us.netrek.org> wrote:
)
)Is linux faster to run? I'm guessing it would be, but I'm not sure.
Of course.
-Tom
Ra\/en
Tom Holub wrote in message <6epc52$rht$1...@shell3.ba.best.com>...
You mean, "Of course, since Microsoft didn't make it."
Steve Sheldon
Shawn K. Quinn - NO SOLICITING
In message <6ep6n4$pkf$1...@shell3.ba.best.com>, Tom Holub
| In article <6eorgk$1e0$1...@news1.bu.edu>, Alec Habig <ha...@budoe.bu.edu> wrote:
| )
| )The message to all you people whining about it is this - download the
| )source and put some work into it. Prove me wrong.
|
| Or install linux.
Or even FreeBSD.
--
Shawn K. Quinn - skq...@brokersys.com - please consider visiting my
home page at http://purl.oclc.org/net/skquinn/personal
Geek code listed
Mike Denison
>Is linux faster to run? I'm guessing it would be, but I'm not sure.
Is question more to make sense?
--
na\/el
Trent Piepho
>>In article <6eorgk$1e0$1...@news1.bu.edu>, Alec Habig <ha...@budoe.bu.edu>
>wrote:
>>)
>>)The message to all you people whining about it is this - download the
>>)source and put some work into it. Prove me wrong.
>>
>>Or install linux.
>
>Is linux faster to run? I'm guessing it would be, but I'm not sure.
We did some tests when I was in the dorm with ethernet. Machines running
linux consistantly had less lag than those running windoze, with the same
hardware. This was with ethernet cards, so lag with modems and PPP may be
different. We measured lag with the netrek server, assuming that the code
would be the same that way.
--
|Gazing up to the breeze of the heavens \ on a quest, meaning, reason |
|came to be, how it begun \ all alone in the family of the sun |
|curiosity teasing everyone \ on our home, third stone from the sun. |
|Trent Piepho (xy...@u.washington.edu) -- Metallica |
Darryl Palmer
Alec Habig <ha...@budoe.bu.edu> wrote in article
<6eoq2m$lt$1...@news1.bu.edu>...
> Forget 95, you really need a multitasking operating system to do a
> server. Context switches are the big server bottleneck.
You could do the server in 95, yes NT is more efficient but 95 will work
also. The big problem with doing a server on 95 or on Windows NT
workstation is that you are limited to 10 simultaneous inbound connections
in the license agreement, the same clause that Microsoft used to stop
Netscape from marketing their Internet server that would support 100+
simultaneous connections on Windows NT workstation.
Darryl Palmer
Darryl_P...@msn.com
gdea...@netset.com
In article <6eorgk$1e0$1...@news1.bu.edu>,
ha...@budoe.bu.edu (Alec Habig) wrote:
> Programming clients and servers takes both a fair bit of know-how and a
> desire to tinker with source code (as well as spare time to burn).
>
> People who know what they're doing and have an itch to mess with the way
> programs work aren't using Windows for both reasons.
I was considering this exact problem a few weeks ago during the thread
about developing a new Win95/NT client, an idea of which I approve but
probably wouldn't care to work on. A neat solution that came to mind was
the OpenStep/Yellow Box API.
Apple intends to release a Yellow Box for Rhapsody (of course), MacOS,
WinNT, Win95 and misc commercial unices such as SGI Irix. Additionally, the
GnuStep project is creating a free version of the API for Linux/UNIX.
Ok, so what does this mean? Well, ideally, you should be able to write a
YB-compliant program (read: netrek client) and have it compile on all the
above platforms with minimal to no tweaking. Developers interested in
different platforms could spend their efforts improving the featureset
of this 'unified' client rather than trying to make it compile or look
pretty on foo. The end user gets a client on his favorite "well, it
came with my computer" OS without making the authors share too much
in his misery. Considering the ongoing attrition of Netrek hackers, I
think some consolidation of effort would be a Good Thing.
Of course, the idea still has a lot of problems and unknowns. The big one is
that nobody knows what direction Apple's going to turn this quarter, much less
by the time I could get around to hacking on this...just ask a Newton
developer. They've already downsized the scale of the Rhapsody project by
positioning it as a server-only OS, which is somewhat less than encouraging.
Also, GnuStep is making impressive progress, but is still far from a usable
release.
Anyways, I should be getting Rhapsody in a couple of months, and some
free time about another month after that. If everything holds together,
I'll probably take a crack at it. My secondary interest (aside from the
port itself) would be to GUI-fy the client, making all options selectable
through menubar and dialog boxes that today's spoon-fed gamers can deal
with.
Any comments or critiques welcome. And if the formatting of this post turns
up screwy, my apologies. I'm posting through Netscape/DejaNews for the first
time.
-Greg Dearing
-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/ Now offering spam-free web-based newsreading
Hareendra Yalamanchili
> > >Does anyone know where one would find a server for Win 95 or NT?
> Yes, well, the code is truly ugly.. It in fact would be quite
> difficult to compile it under Win 95/NT.. Maybe using Cygnus tools you
> could do it..
> Hey, perhaps someone could rewrite it? :)
Has anyone tried using the EMX runtime librares to port to OS/2?
Hareendra Yalamanchili
hyal...@mit.edu