Mijzelf

My MSDN Library April 2001 says:

Alas MS dumped all information about older window versions. My MSDN Library April 2001 says about SHDeleteKey:

The IPv4 address space can handle 4,294,967,296 addresses. Do you really think a single desktop computer will ever connect to that many unique addresses in its lifetime, let alone within the period of time such addresses would need to be cached? Depending on the implementation of the mapping table there are 'only' 16 million IPv4 adresses to burn. When transparantly implementing the ipv4-in-ipv6 encapsulation, you can only use a (or all) private range, because the other addresses can still be used, outside. Of course it's possible to just map the outside ipv4 address to another ipv4 address inside, (which will not give the full address space, you cannot use the reserved ranges), but it will break software which uses hardcoded ipv4 outside addresses. (Or softcoded, like manually configured DNS servers). Can a single desktop computer burn 16 million IPv4 addresses in a few hours? I don't know. When you search the forums you will find a lot of people suffering from instable internet because their router cannot handle the hundreds of connections generated by p2p networking. Further you can read about plans to distribute (hired) movies via p2p, and maybe even stream via p2p. Nobody could have predicted that a normal consumer now can have several hundreds of connections at a given moment. I wouldn't be surprised if in the near future the same costumer will be burning 16 million addresses in a few hours.

Maybe you can use Wireshark to find out what it's doing.

You're worrying too much!!! The simplest solution is to use a router to translate IPv6 to IPv4. It just works! dencorso, do you have actually seen a router work which does ipv6 on wan and ipv4 on lan? I think it's very difficult, maybe impossible, to implement. The problem is the size of the routing table, and the number of available ipv4 addresses. When a big hosting server changes to ipv6, it doesn't have to use virtual servers anymore. Instead it can assign all it's domains a unique ip address. Which means, worst case, that the router has to map an ipv4 address on each domain visited, and remember this mapping virtually forever, but at least for a few hours, because it doesn't know if it's cached somewhere in the lan. So the mapping table can grow *very* big, and it can even hit the limit of available ip4 mappings, because there are more ipv6 addresses outside than mappable ipv4 addresses inside. This could solved by running a http proxy server on the router, but this works only for http.

Maybe you can get more information when you disable the bootlogo. Then you can see the BIOS and DOS messages while it boots. In order to do so you have to edit c:\msdos.sys (a textfile with flags hidden and system), and change Logo=1 to Logo=0.More on editing msdos.sys here.

That's a good thing. The only way I'm aware of is just entering the url. When it doesn't exist TWM offers to search the whole domain, but that never gave me something useful. Given one piece of hay is about 1cm3, this would give a haystack of 150000m3. A very big haystack indeed. (Assuming it's an American billion. A European haystack would even be a thousand times bigger)

What size are we talking about? I remember there is some limitation on a HBITMAP size in 9x (16MB? Alas MS stripped all 9x information from the online msdn documentation, and at the moment I've no access to my offline files)

You always want to go to the bottom, don't you? When you stumble on a dead link, you can try the Wayback Machine, and, in this case, it worked. In this case, I think you can determine *if* there is a resource leak, by using the resource meter tool, and scan again.

That sounds like a resource leak. I suppose that somehow the combination virusscanner/NetBEUI somehow leaks resources. In that case the problem is on the W98 side. What kind of malware are you afraid of? I can hardly imagine a Wfw box can be infected through the internet these days. Malware uses browser bugs, and/or user bugs to install 32 bit software. But I don't think it will run on Wfw, even when it has win32s installed.

Do the problem images all have the same color depth? To print an image a conversion has to be done from RGB to CYMK. Maybe the conversion fails for one particular color depth. If that is true, the reason that not all applications have this problem could be because some programs convert all images to the same color depth for internal use.

When IBM inserted a harddisk in their PC, back in '80 or so, they designed a partition table which could contain 4 partitions. Primary partitions, nobody had heard yet of extended partitions. At some point somebody thought 4 partitions is not enough, and he invented the extended partition, which is nothing more than partition with a partitiontable in it, serving room for 4 sub-partitions, logical partitions. So you can create 4 primary partitions, or 4 extended partition each containing 4 logical partition, and you can even chain extended partitions, instead of a logical partition you can add another extended partition. In normal use it doesn't matter if you use primary or logical partitions, but in most cases you can only boot from a primary partition, and Win9x first assigns driveletters to all primary partitions, and then all logical partitions. Then there are software limitations: fdisk can only create 1 primary and 1 extended partition (filled with max 4 logicals). Microsoft states that using more than one primary partition on a harddisk is not supported on W9x, and can lead to unpredictable results. (I've never noticed any problems) So when you want to add a partition using fdisk, your only choice is adding a logical one. When you use Ranish, or another PM, you can decide to add a primary. MS won't support this, but hey, they support nothing at all, when you use W9x. In both cases it will work fine.

No, sorry, I meant the tabcontrol.

You'll need an adapter to connect it to a desktop computer. While it's electronically the same connection, a laptop disk has a smaller connector (smaller pitch). About jumpering, when you connect to a cable of it's own, you don't need jumpering. When your laptop can boot from CD, you could use a gparted live cd to add the partition. That is no problem. Any partition tool will do, as long as you don't touch the current partition, and only add a new one. Actually, I think dos+fdisk of W98FE is the same as dos+fdisk of W98SE. That will also work. I'm not sure if you can add an overlay without having to repartition the disk, but if rloew thinks his overlay will work for you, I assume it's OK.

By loading them in config.sys.DEVICE=C:\Path\to\driver.sysNormally very few drivers are loaded. Things like sound are to be supported by the programs using them. Yes. You'll have to backup your MBR, and put it back afterwards. Further you'll have to backup your dos mbr, because you'll need it to get the XP loader to load dos. Can be done, but it will cost an awful amount of memory. Here you can find some information about creating a (dos) bootdisk with network support.

8GB is the maximum disksize using CHS addressing. So apparently your BIOS doesn't support LBA addressing. Windows uses it's own drivers, so it shouldn't have problems to access the whole disk. The problems you are facing are: - The partition tool of Windows (fdisk) is a DOS program, which uses the BIOS for disk detection and access. So fdisk can't create partitions beyond 8.4GB. - Windows is loaded by DOS, so all files needed to start Windows need to be in the first 8.4GB. The first problem can be solved as easy as connecting the disk to a more modern PC, and add a partition filling the rest of the disk. (Maybe you can even use fdisk when running in a DOS-box under Windows). The second is solved by discipline. Never install anything on the new partition which is loaded early in the boot process. I once made a W95 system unbootable by installing a virusscanner on the wrong partition. Obviously a part of the scanner is loaded before the Windows disk drivers are loaded.

You can simply subclass the tabcontrol to get that message. LRESULT (CALLBACK* OldWndProc)(HWND, UINT, WPARAM, LPARAM); LRESULT CALLBACK SubClassFunc( HWND hWnd,UINT Message, WPARAM wParam, LPARAM lParam ) { if ( Message == WM_CONTENTS_CHANGED ) { // Whatever } return CallWindowProc( OldWndProc, hWnd, Message, wParam, lParam); } hWnd = CreateWindowEx(0, "RichEdit20A", ... ); // Now subclass the window that was just created. OldWndProc = (WNDPROC)SetWindowLong( hWnd, GWL_WNDPROC, (DWORD) SubClassFunc );

Looking though your code, I see that your tabcontrol fills the whole clientarea of it's parent. Further it seems your richedit control has the same parent as the tabcontrol. If that is all true your should make the tabcontrol parent of the richedit control. Further there is a bug in this code: HFONT fontHandle = CreateFont(lfHeight, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, "FixedSys"); if (fontHandle) { SendMessage(mHandle, WM_SETFONT, (WPARAM)fontHandle, MAKELPARAM(FALSE, 0)); DeleteObject(fontHandle); }You shouldn't delete the font until mHandle doesn't need it anymore. That is when mHandle gets another font, or when mHandle is destroyed.

I think you just have to call UpdateWindow() after creating the control.

@dencorso: Doesn't using such a giant hosts file give a serious performance hit in name resolving actions? I don't think the name resolving system on W9x is designed to cope with such a database.

No, that is called a carefully written service. For instance I'm running lighttpd (a webserver). Literally thousands of people have been examining the sourcecode, and the last known vulnerabilities have been fixed in 2007. So I dare to expose that server to the internet. Btw, a firewall won't add any security. You can compare a router with a telephone exchange. An open port is then an extension number. Without extension numbers it's impossible to call a particular phone behind the exchange. But all phones can call out. When a port is open a peddler could call and try to sell the service (your daughter) an ipod. Maybe it's not a good idea to open that port. A firewall could be a telephonist, which decides whether or not to connect in incoming call to your daughter. A peddler will never reach your daughter, but her friends can call freely. On the other hand, when a friend turns out to be a peddler, you'll have a drain in your bank account. In that case you'll need a 'deep packet inspecting firewall', ie a telephonist which listens to the conversation, and pulls the plug when something goes wrong. When your daughter is called by her Chinese friend, the 'deep packet inspecting firewall' won't work because it doesn't understand Chinese. Because there are many languages a firewall can't do much to protect a weak service when a connection is already made. It can only listen to the conversation (which costs *lots* of CPU power) and hope it will recognize it when the conversation becomes evil. (And hope it's not a false positive). So you should only forward calls to a bullet-proof service. Your mother-in-law. No way anybody could sell an ipod to her. Your Westell is fine. Summarizing: - All NAT routers have the same inbound protection, which is strong. - A firewall in a consumer router can give hardly any protection. So another router would not add any safety. It could only give you more options to arrange your network, but seeing your questions I don't think your are waiting for more options. You already have a network, in which the router is hardly involved. The router only assigns IP addresses to all PC's, and after that all traffic between the PC's is handled by the switch.

This firewall seems pretty useless to me. High and Medium will block all outbound traffic, except the most basic services. You can surf the internet as long as the webpages are on default (http,https) ports, but for instance streaming video won't work. Minimum is the same as No, except that is protects against 'known attacks', whatever that may be. The only possibly useful option is Custom, depending on how configurable it is, but seeing the rest of the "firewall", I'm not very hopeful. Wrong. The services are bullet-proof because they are not vulnerable to attacks. Even if a hacker knows which services I run on which ports (which is partly not difficult to find, a simple portscan will show I'm running a webserver, the other services are using non-default ports), he can't do anything with it. For the weaker services I trust my router not to expose them, because I didn't forward any ports to them.

Can't say anything about this particular box, as I haven't seen the manual and/or specs. And yes, I have had a ISP provided modem/router which supported (some of) these functions. That was a Copperjet 801 when I remember well. (BTW, it had a single UTP port, and the box was configured in bridge mode). Well, let's say I have forwarded some port, and the firewall detects a portscan, and closes all ports for a few minutes. Your service is unreachable for the same amount of time, but the scriptkiddie on the other site has not found your open port. When your typical user doesn't open any ports, he probably won't notice the extra firewall functionality. But I think lots of people *do* open ports. It is needed for many games and for torrents. Googling on 'port forwarding problems' gives 2,290,000 hits.

You can skip that 'kind of'. A router is a router. When you only do some surfing it's hardly useful to have a hardware firewall. As soon as you start exposing some services (by portforwarding) to the internet a portscan detector could be useful, but I don't know if it's provided. I found that manual for you, but I didn't want to register to be able to download it. So I don't know which functionality the firewall has. I never bothered to configure the firewall in my router. I just trust my exposed services to be bullet-proof.