Ninho

Now BenoitRen you're seen battling over pin points, I won't (even try to) dispute your clicking habits; the important is you've seemingly backed off your previous arrogant and dangerous assertion, that the windows desktop update should never be applied. I'm happily done with this (very off) topic, and I only hope you won't again change your mind and jump in with that silly idée fixe the next time anyone in passing mentions the desktop update. OF course, I do not deny your absolute right to prefer the old Explorer look aznd feel and to keep it for your own use, however bizarre it may seem... Just don't tell people it's the right thing for them 'cause it isn't, definitely. -- N.

BenoitRen said : (Sigh!) I have zero hate. But your saying and repeating such things as : ... makes you look like a Troll, and I have no time or willingness to feed the Trolls ;=) Alright let's assume you are being earnest, and let's review your points : you dislike - and rightly so - the web integration, "active" desktop, web folder view, channels, etc. True, Microsoft has most of that obtrusive bloat enabled by default But what you don't seem to realise, each annoying feature can - and IMO should, be disabled ... Nope! No hacking, not even geeky tweaking, just plain settings, courtesy Microsoft. Methinks you are the hater here : you hate the default look of "active" desktop so much you haven't even tried to taylor it to you needs, which however only takes a few minutes and is accessible to the basic user, the more so to a geek like you. Exercise : Just try it when you have an opportunity again someday ! After pouring some sweat maybe you'll be charmed, because the newer desktop does bring better UI elements, some conspicuous like the Quick Launch Bar, some subtle but nonetheless noticeable like much better drag and drop functionality to the Start Menu... (Note : I'd be suprised anybody hated Quick Launch, but in case... you could nuke it, too, easily) .... (snip...) With Explorer as the shell, yep, about 20 MB here; with WinFile instead (use FMLfns : www.wincorner.com ) : 14.5 MB. Try it for yourself ! Later, -- Ninho

BenoitRen : shouldn't you add "IMHO" to your strongly worded (and mostly unfounded) opinions ? The elements you mention (web integration, web on the desktop, browsing HD as a website) are all optional (and I never used any of that). OTOH there come many useful UI elements with the IE 4.0/ Windows 98 style desktop update. Plus, should you never install it, you'll miss many security updates. In any case, had you *read* (and understood) my own humble post, you'd realise that what I'm proposing in this thread makes no use of Explorer as a shell, hence no "desktop" with or without web integration, no icons on the screen background, no memory bloat and no nightmares. You get something like the old Windows 3.1, just much better. As a test, I've now setup my 486 DX-2 with 32 Mbytes main mem, Win 95B, per my post above; less than 15 Megabytes of RAM and /zero/ bytes swap file are in use. I'm not hoping to convince you yet. This is unfortunately not the first time I've noted to myself the vanity and vacuity of most of your misleading comments on this forum, recently it was apropos that animated cursor vulnerability IIRC. This is why I carefully avoided to quote you in my above post, and shall not reply or mention you any more. Sorry I've gone out of my road on this occasion. Regards -- Ninho

BenoitRen : shouldn't you add "IMHO" to your strongly worded (and mostly unfounded) opinions ? The elements you mention (web integration, web on the desktop, browsing HD as a website) are all optional (and I never used any of that). OTOH there come many useful UI elements with the IE 4.0/ Windows 98 style desktop update. Plus, should you never install it, you'll miss many security updates. In any case, had you *read* (and understood) my own humble post, you'd realise that what I'm proposing in this thread makes no use of Explorer as a shell, hence no "desktop" with or without web integration, no icons on the screen background, no memory bloat and no nightmares. You get something like the old Windows 3.1, just much better. As a test, I've now setup my 486 DX-2 with 32 Mbytes main mem, Win 95B, per my post above; less than 15 Megabytes of RAM and /zero/ bytes swap file are in use. I'm not hoping to convince you yet. This is unfortunately not the first time I've noted to myself the vanity and vacuity of most of your misleading comments on this forum, recently it was apropos that animated cursor vulnerability IIRC. This is why I carefully avoided to quote you in my above post, and shall not reply or mention you any more. Sorry I've gone out of my road on this occasion. Regards -- Ninho

The desktop update should be applied, it refreshes many files and generally makes for a better UI. Plus, conformant to MS ways, they ceased correcting or testing the older explorer against newly discovered vulnerabilities. Whether or not you have the desktop update, however, nobody is forced to use Explorer as the Windows shell ! Windows 9x can run happily without having a single instance of Explorer/IE. The key here is to choose your shell accordingly (the "shell= ..." line in the System.ini file). On a system with little memory, I do use and recommend the good old file manager (Winfile.exe) as the Windows shell. Well , not quite : I see you shuddering, what cr*p! Winfile doesn't "do" long filenames, does it ? Yes it does, thanks to a tiny shareware called FmLfns (browse for the little jewel!) Having the following line inside of System.ini : shell= fmlfns will in fact launch Windows with Winfile as the shell, including long filename support. You could arrange 2 windows inside Winfile, tiled horizontally, one dispaying the Start Menu folder and one displaying your old Desktop & you can launch most of your programs from there ! For other needs, Control-Escape (or the *Win* key, or double-clicking the background) will open TaskMan. Taskman's *File* menu gives access to a Run app... box, as well as to the various Windows shutdown options. For a browser, you might try OffbyOne (of course IE will work if it has to For a clock, try the round, analog clock which came with the Power Toys. Note : the shell=... can be anything Windows can execute, incuding a script or batch file which will launch the file manager (Fmlfns above) plus your round clock and whatever programs you want to autorun. Contrary to popular belief, Windows does *not* close itself when the designated "shell" is terminated - unless it like Explorer is designed to initiate Windows shutdown if it recognises that it was launched as the shell and it is being closed. Hope this help -- Ninho

Note : The unofficial fix also works in Windows 95 ! Cf. : http://www.msfn.org/board/index.php?showto...st&p=643507 -- Ninho

please feel free to test it on your 95 computer, and if it works ok, I'll modify the installer to include all Win95 editions. I am pleased to report that the unofficial U891711 seems to be working perfectly in Windows 95 ( OSR2 with IE 5.5 SP1 here); of course the installer will refuse to start, so I extracted manually the KB891711.EXE & Q891711.DLL to (my Windir)\System\U891711\ ... and under the HKLM\...\RunServices registry key, I added name U891711, value : (myWindir)\SYSTEM\U891711\KB891711.EXE . [Disclaimer : Reader! don't try the above unless you feel confident you understand the necessary steps and the way to undo them, if needed] After a reboot the proof of concept malformed animated cursors do not crash Explorer any more. Thank the anonymous autor of this patch, and please MDGx will you review your installer to allow for Win95. -- Ninho

Sure, but in the Win 9x series, it is USER.EXE which needs a patch. USER32 is just a small stub, all the meat is in the 16 bit USER. However, the tiltle of this thread is misleading : in Win 9x, the malformed ani exploit does lead to a GPF while some internal USER function is trying to return to garbage, the GPF is caught by Windows which makes the calling process crash (generally, it'll be the Explorer.exe shell which is then auto-restarted by Windows). So this is at most a "denial of service", especially if you had the bad .ani file lying on the Desktop - this being part of explorer, causing an almost unendable chain of crash/restart... *BUT* contrary to the Windows NT/2k/XP... series, on Windows 9x in no case can a "sploit" of this kind cause instructions, contained as data in the malicious file, to be handed control and executed. This is immense superiority of the Intel X86 *segmented* model over the (easy to use but lame) "flat" programming model adopted by MS in NT and ff., and also, regrettably, adopted by Linux even on the X86. (Only the first version of OS/2 had it right. Twas a Microsoft product by the way, which shows MS could do things right if they wanted, but did not - money not right-doing being their goal ). And, oh! yes, I've done some "debugging" (soft-iceing...) of the ani crashing explorer before posting this answer. Cheers, pals

It is indeed possible, but generally not worth doing unless, maybe, your system has undergone many changes since it was first build. Working instructions (by Pietrek or Russinovich or? some guru of the kind) are available on the web should you want to play, Google is your friend... [Edited : The guru in question was Clive Turvey indeed, as referenced by Jacklaz while I was writing this...] Basically, vmm.vxd is a composite of : - "the" VMM.VXD itself, - an "W4" archive of other VxDs, similar to a zip, built at system installation time from the individual VxDs. Because the VxDs are compressed and also since it is only one file to open instead of several, the load time is indeed decreased. Notice that this was already done in Windows 3.11 for Workgroups, the Win386.exe was also an archive, albeit uncompressed (the "W3" type). HTH -- Ninho

A word of warning about one of these "tips" : It should be noted that using the Win ME version of Defrag in place of the 95/98/SE versions, as suggested in this thread and elsewhere too - is *NO* option on disks which use MS DoubleSpace / DriveSpace compression. (For all I know, DxxeSpace was removed from the crippled version of Windows akaWindows ME.) All it can do is defrag' the underlying uncompressed drive, which is generally moot anyway ,...but the code for defragging the compressed volumes is simply not in there :-( HTH -- Ninho

An excellent question, Ctrl-X! I would post the output of a "route print" from Aux and from Main - but I am not @ that location any more. Whatever, the routes table on Aux is set up automagically by the Windows 95 "direct parallel connection" launcher. It doesn't have a specific route to the 10.* network, but then the default route (0.0.0.0) which has Main as the gateway should cover the case anyway. As for return routes to my /24 (could even be /31 !), nope, routers/computers on the 10.* network do not know of their existence, and I do not have the authority/possibility to modify this state of affairs; however it was my understanding that - assuming my Main station itself were routing correctly - return routes to the 192.168.55.* would be automatically and dynamically recorded by the routers involved as these would see outgoing traffic sourced from my Main's MAC address and a yet unknown IP. Isn't this how internetworking with IP works ? Realising I may well be making a fool of myself :-) Since AFAIK it is not possible to persuade the Windows parallel networking module to use IP addresses from the 10/8 space (nor any IPs other than 192.168.55.x) then I guess it will be very complicated at best to achieve what I was dreaming of aloud. Thank you very much for your help. And I must add it is not important if the above cannot be done... as long as I have learnt something new. -- Ninho

Begging for help from the Gurus before I've pulled all my hair out ! Brief description : - "Main" is a Win 2k Pro station connected to a ethernet LAN and part of a domain. - "Aux" is a standalone old comp running Windows 95 B. I have Aux & Main connected & networked point to point thru the Parallel Cable, thus making a 2 stations mini-net. The parallel connexion uses TCP/IP as well as NetBEUI. I'll be concerned with TCP/IP only however. For reference, "Main" is assigned IP 192.168.55.1 while "Aux" has 192.168.55.2 The local "mini net (over parallel cable)" works as intended. Main is pingable from Aux and vice-versa, TCP and UDP between them work as well, for instance I can browse the Internet from Aux thru a local HTTP proxy running on Main (privoxy). But what I can't achieve is any IP connectivity from/to Aux to/from LAN THROUGH the Main station ! IOW, "Main" doesn't accept to be routing IP between the parallel cable interface (IP adresses 192.168.55.*) and the larger LAN (IP adresses 10.*.*.*). From Aux, the one address in the 10.* range that answers "pings" is "Main" itself. No IP traffic goes further AFAICT :=( Now wait before jumping to conclusions : I do know Win 2k Pro is not supposed to do routing, *except* there is a certain registry setting in : HKLM\System\CurrentControlSet\Services\TCPIP\Parameters namely, IPEnableRouter , which I set to (DW) = 1 and that, according to what is found at different places on the net - and at Microsoft's own technet - should make the Win2Pro route between all its interfaces. Unfortunately even with that setting, the observed behaviour is as described above :=( So, MM. Gurus, what is going wrong ? Of course Windows is flaky when compared to say, Linux, as far as networking is concerned. Is Win 2k incapable of routing between the Direct Parallel and Ethernet adapters ? Please tell me if it's a no-no, as I've spent so much time on this apparently simple thing without a hint what's going wrong. -- Ninho

LLXX wrote : You couldn't be more wrong. VxDs can contain both 16-bit realmode *and* 32-bit protmode code in different sections. Yes, that is why the LE/LX format was brought to Windows (from OS2). Indeed I was coming here to amend my statement so as to state what you say above clear. I erred yesterday, my feeble excuse is I haven't put my hands into dirty Windows 3/9x system code since ca 2000. Again, believe me or not, I disassembled lots of vxd code and assembled mine in turn, *and* of course I studied the DDK and VMM functions. Not a pleasant sight altogether ;=) @Petr: the best way to 'see' vxd code in action is with the help of a kernel debugger, SoftICE, TRW or Microsoft's own. There used to be a specialised disassembler for LE/LX, by an author who also wrote a PE disassembler. Edit : I'm being an old fool again, reading the thread shows you must have got all the toolz & what you were really after was the LE file layout ;=) Best wishes for success ! -- Ninho

Petr, you are misguided, vxd's are 16-bit code. You are excused : I see this misconception all the time (including in the Wikipedia. Don't believe everything you read on the 'net:) Basically, Windows 9x is a 16-bit OS which runs 32-bit apps. The system code is roughly half 16-bit, half 32-bit. The virtual device drivers (vxd's) are 16-bit code (of course, they can and do manipulate 32-bit DATA, but that's another story). This is why they use the LE format, not PE ! This is also why your disassembler shows you garbage. You need to instruct the disassembler to show the code as 16-bit, or use another disassembler. Yes, I have written quite a few vxd's myself (in MASM)... HTH -- Ninho

Hiyya, Forumers ! I've had to put an old box with Windows 95 SR2 to active service back. Now seing MS - disgustingly - have pulled most of their updates offline, as occurred to me when trying to download the MFC libraries update yesterday - is there a place where I could still safely download missing updates from ? First choice will be a site with downloadable updates for national editions (the French one in my case). Of course, I will reluctantly make do with the US updates if necessary. And, is there anything like the ultimate 95' update plan ? See you -- Ninho

Erpdude8, unfortunately that page has inaccurate and misleading information, namely where it states : Not so - as I stated (and others have confirmed) in this thread, SiS only provides hiding of the USB 2.0 capability from Windows for mostly cosmetic reasons. I have since abandoned all hope of making the SiS 741 based USB 2.0 work in Windows 98 SE (or ME). If I be proved wrong so much the better... but don't hold your breath ;=) Regards -- Ninho

I've never used command.com replacements, so I can't test NDOS or 4DOS command interpreters. But I have added [98SE2ME 3-19-2006 edition] detection code into all 98SE2ME batch files for COMSPEC and COMMAND.COM, and if none are found, then the environment will be set to the current location of command.com, accounting for all 3 possible locations: C:\ root, %windir% + %windir%\command Thanks for looking at the question, MDGX ! I believe setting the COMSPEC to point to a copy command.com will not work however : you'd need to spawn an instance of that command.com and have it exectute your batch, in addition. This may be unnecessarily complicated, what I suggested was only : /check/ the current command interpreter is MS Command.com, and if not, bail out with a polite explanation. A first check which will immediately single out 4DOS/NDOS (but not other non MS DOS Commands, if such exist in the real world) is the following scheme : IF '%@EVAL[2+2]'=='4' GOTO 4DOS HTH

Unfortunately if not unexpectedly it's always the same SiS scam : disable USB 2.0 in Win 98/SE/ME. Was worth trying anyway...

No need to be sorry... But I have tried all you are suggesting, as reported in the other thread. The fact is, the OrangeWare do not want to hook any device, even after nuking & reinstalling all in the proper order. It turns out the OrangeWare drivers were never intended to be freely used by us mere mortals as a replacement for non existing MS drivers, instead they are meant to work only with approved hardware (read: paid license). They'll work mostly with add-on USB 2 on PCI adapters. The hardware ID of supported ware being encoded somewhere inside the driver code AIIUI. If I understand clearly what you have got is such an add-on USB 2 card with, probably, a NEC chip and this is supported by OrangeWare. I suppose I could buy such an adapter if I needed USB 2 on Win 98 SE badly...

Your file, WIN98.7Z , seems to contain only a WIN2K foder. Did you not forget to include the 98/ME part, or am I missing something ?

Ack! Could give it a try... ProblemChyld, could you in addition please post a link to - or mail - the accompanying three usb*sys files, in case they are different from their Win 2k counterparts ? Cheers, -- Ninho (mail to ninho _at_ infonie _dot_ fr )

Hi again ! This msg poses two questions, astutely labelled 1. & 2; : 1. Using the win 2000 inf : no way... I've just checked again - I remember now, I must have tried it before, just after I had built this machine - and definitely NO, the AsRock/SiS provided drivers intended for Windows 2k will not install on 98 SE. There are at least 2 problems, one trivial : the "inf" file is unicode and Win 98 will not even try to read it I believe, and 2. even if it were converted to 8-bit text, it is 2k only, would need serious revision by one of our competent fellows. One could copy the USBEHCI.SYS, USBHUB20.SYS and USBPORT.SYS of win 2k to the DRIVERS folder of windows 98 SE/ME and there is a faint possibility they might work PROVIDED, of course, a tissue of relations was properly established in the Registry base. Which in part is what the "inf" are for... Can someone with experience in "inf" files and "sys" WDM driver on '98 look into this ? I know my way with old style Windows "vxd" drivers, I even wrote a few of these (in assembler!) but I have no idea and no docs whatsoever about the WDM... :-( Or by chance there is a working on 98 "inf" file somewhere which could serve at least as a model ? Among us are people with, I believe, access to many brands of computers & motherboards. It would be interesting to look at ALL the modern mobos with SiS chipsets. 2. Orangeware trial driver for developpers (cf. Petr) : The link provided to the expiring trial USB 2 driver by OrangeWare is password protected. Could we be lucky enough to find a non-protected download, for evaluation (cough) ?

A suggestion to MDGx : in your batches you could check for the DOS shell being indeed Command.com and not some replacement, lest people get into trouble; while installing a reduced SE2ME for the USB components, I got caught by such a problem with COPY /Y not understood by my shell being NDOS (a 4DOS variation). Fortunately I spotted the problem and since I'm "fluent" in DOS twas a no brainer to correct for it. If you don't want to go to the trouble of checking that Microsoft's "command.com" is in memory and is indeed the active DOS command interpreter, a warning on the screen or at least in the documentation is needed IMHO to the effect that this batch must be run by MS command.com *only*. Regards, -- Ninho

Done this several times & almost routinely. Thank you anyway... There's a thread in another section where I report what I've been doing. Next trial I'll zap the OrangeWare again and see if Windows 98 accepts the SiS driver intended for Windows 2000 (suggested by Eck). We'll see... -