flatten and rebuild---REPLACE!

Please, do so. You really don't need DNS and ICMP, neither HTTPS. ;-D Anyway, with Win95 you're f***ed anyway.

Stupid idea.

Well, what about it? If those guys at AllOfMP3 and eMusic would actually offer what I want, I'd stop legally downloading it for free from P2P networks.

But well, no other effect. Honeypots don't work as a security measure.

I am actually playfully agreeing with your basic philosophy Seb~. I tried to disable DNS and ICMP and....oh, your being sarcastic. Silly me.

Seriously though, I have tried to get my cable ISP to tell me what minimum I require enabled for a access and all i get is ports 80,25,110. Nothing about which protocols, nothing about 'inbound or outbound' ... I have found out that without DNSlookup the process is so slow i might as well get dialup, I also found that i can't refresh my IP without...DCOM? so, by trial and error I creep along on my hands and knees looking for a reasonable solution

I know...I'm so weak.

warf,,,,begs your advice.

You should really consider if this lousy support is worth the money you're paying.

Well, maybe they assume that you should know at least the most important details about the protocols involved if you're asking for such things...

Huh? Without DNS, you won't get any DNS resolving. Are you twisting this with the Windows DNS caching daemon? What about your DNS configuration?

Huh? That's definitely strange.

Well, why don't you take a look at ? An extensive discussion about what ports are used by which daemons and how to configure them.

I know you are being facetious here (no one is crazy enough to trust windoze95 near a network connection, never mind the Internet), but this point suggests a misunderstanding on port numbers and how they are used in the big picture. Think what you put on an envelope when you send mail. You put the address of the destination - and in the Internet, this is found in two locations. The IP address of the destination is the first address (bytes 16 to 19) of the IP header - which sends the packet to the destination _computer_ out "there". But in the TCP header, there is a destination _port_number_ in bytes 2 and 3 to tell the which service on that destination computer to deliver this packet to. See RFC1180 (or RFC0791 and 0793 if you want the actual specifications) for additional details.

But just as you put a "return address" on that letter, the packet also has your IP address (in bytes 12 to 15 of the IP header), and the source port number (in bytes 0 and 1 of the TCP header) where the packet came from on your computer. Point is, the 'source' and 'destination' port numbers are not the same. If they were, what is the point for having both in the header? The server is usually on a "well known port" (in this case, 80), but the client will be on an ephemeral port number (the "next available number) between 1025 and 65535.

As you are not offering services to the world, anyone attempting to connect to a port between 0 and 1023 on your system should get a "No one lives here" answer - which occurs BY DEFAULT when there is nothing on the port. With one exception (DHCP client), there should never be any packet leaving your system with a _source_ port in that range.

Old guy

SERIOUSLY...thanks. I am of course at your mercy as you 'might' be regarding D-orbital ab inito calculations for bis-phenylphosphorylation of....Motrin, or whatever.

As is apparent, I am asking these questions to learn enough to 'play ball' but am also realistic to know [I reiterate] a masters in comp Sci in order to stick it to HP ain;t gonna happen. I would however appreciate a reference to an intermediate treatise....magazine, covering enough of the basics to at least enable me to ask the right questions. At least then I might be able to assist myself. Really, there is a lot of chest pounding in these forums but the chanced to educate the eager is the most noble use of these ephemeral packets these is no? Warf...looking to you[se] for a ray of light.

What about DNS query fallback? If some queries with source port > 1024 fail, some resolvers resort to source port 53. Also quite common behind NAT routers.

Sigh, a lot like like listening to very intelegent men argue specifics of hard science from opposite perspectives; how can two correct people be in disagreement? [as i suspected, no hope for me.] miffed again.

Now that would be a good question. I don't subscribe to magazines like that - they tend to be rather useless for me as everything is aimed at the windoze user level. The more technical magazines tend to expect that the reader has the basics, or is willing to spend the time scanning the details out of RFCs and the like. Do you have access to a good library? One of the better books in "TCP/IP Illustrated, Volume 1" by the late W. Richard Stevens (Addison Wesley, ISBN 0-201-63346-9, 1994 [there is a 1996 edition as well], 576 pages, US$lots) that is normally used as a text book in college networking courses. I would NOT recommend buying it in this situation, but if you can borrow a copy, it may be worth the read.

As for the way an application communicates, other than the simple overview (in Chapter 1 of the Stevens book) this tends to be more O/S specific, and as I've stated, I don't do windoze.

Old guy

snip....>

SIGH....... Warf.