Drugwash

Any kind of driver that still supports Win9x should be distributed and backed up by us 9x users, for we never know what use it could be put to, in time. Could you possibly offer a valid download link for that driver package (and any software application)?

Thank you for testing and confirmation. I noticed the 7-zip compatibility issue too - I use an assortment of packer plug-ins for Total Commander and the one for 7-zip created by Ono (Adam Strzelecki) is very old (last updated in 2007, newer versions are Unicode only), incompatible with Leyok's packages. The standalone 7-zip (I have 9.25 alpha currently) works fine though.

No problem, thanks for testing! Have you by chance tried it with uSP3 installed? At this point, it's either a conflict with an updated system component, or a hardware/driver incompatiblity. I know the SiS chipsets are quite lousy, but this would be a little bit too much. Sometime these days I'll see if I can put together another machine for testing, with different hardware. However, if this turns out to be a hardware/driver issue, it better get fixed, as there may be others out there with similar setup and thus similar problems. Anybody else tested my latest builds? Anybody else having similar problems with any builds at all (Leyok's, coolman's, mine)?

Yeah well, as long as there's mixed information, it better be as condensed as possible. We don't like 9x haters, so let's not be XP haters ourselves. Thanks for the links!

Here's a package with both builds, from code pulled right after this commit: Download With this build, upon first reboot after installation, I was again able to access the test machine through network, even though it was hung as always showing only the wallpaper and mouse cursor. This time Revolutions Pack 9.7.2 was installed, as I wanted to test if there's any change in behavior (I also have a video issue with 32bit transparency that I wanted to check). Here's the KernelEx log for the Debug build linked above:

If I recall correctly, there's the RD (remove directory) command in NT-based Windows as a counterpart for 9x's DELTREE. I'm not sure, however, if ti works for non-empty directories or what are the accepted parameters, if any. XP (and probably later versions too) can automatically run batch files at startup and upon shutdown. Start > Run > gpedit.msc > Enter/OK will open up the Group Policy editor. Computer configuration > Windows settings > Scripts (Startup/Shutdown) Add the batch scripts to be run on startup and/or shutdown, respectively.

It's 595 lines - none more, none less. I always delete the whole existing folder and then unpack the one from the zip (can't remember how many times I said that already). I never use Explorer and I know how to manipulate files and folders. Files are always deployed by the installer i build. A few posts up there's a package with both installers, you said you'd check them out. All files are deployed correctly - simply replacing kexbases.dll makes everything work. Virtual machine is too complicated - I have spare hardware to test with, why waste HDD space? I can get the log in Safe Mode, but it's useless, it always says the same thing, over and over again. Just the line number is changed, it's 298 now, not 296 as it was previously. Maybe the code needs rebased, maybe you have some particular settings in Visual Studio or in the project, that don't get propagated in the GitHub. I don't know, but fact is everything I build works, except for KernelEx. I've just built Miranda IM 0.10.12.0 this morning, both ANSI and Unicode, on my 98SE machine and the Unicode version works without problems on the XP machine while the ANSI is working on the test machine. That's pretty much code there, if there was something wrong with my setup I'd get errors, crashes, whatever.

Up until now I was able to access the test machine through local network even when it was stuck at boot time; after later commits (zip package downloaded on April 5th at about 1:40 UTC) I can no longer access the machine, although it appears in the Network Neighborhood list. And yes, it still hangs at boot, but I cannot get the log directly anymore.

Hex-editing is easy anyway, but here we have a little more than that, namely adding/removing/modifying resources, such as controls in dialogs, strings and even icons, bitmaps and other, for those wanting a modern look. None of the existing tools I know does the perfect job - not even M$'s Visual Studio 6. So we need a new, specially designed tool. I started to build one, but left it aside, as I got sidetracked by other issues - who knows when (or even if) I manage to pull through. If anybody knows of such existing tool that can specifically work with 16-bit .cpl files and is Win9x compatible, other than the ones already mentioned throughout this topic, please speak (or send PM).

Better yet - here's a package with both Debug and Release installers: download EDIT: After building latest code and replacing kexbases with the one in Leyok's package, I noticed Total Commander 7 loses the drag'n'drop ability between panels. Disabling KernelEx for its executable, fixes the issue.

This was fixed [...]I found the shut down problem [...] Fixed it and logoff/shutdown is working properly. Thank you! I can build it. I can build all parts of the KernelEx project. I always could, albeit with slight changes to project files, at first; now, there's no need for any changes, it builds right away.So I unzip the sources to a new folder, I launch the main .dsw project, I choose Batch build and in a few minutes all subprojects are compiled. I launch the NSIS script twice from context menu (remember I added the Debug item to the registry so I'm making a few slight changes to the .nsi file prior to launching) and the Debug/Release installers are ready. But no matter what, the kexbases.dll built on either of my two machines (Pentium III Win98SE and Pentium4 WinXP-SP3) hangs the boot procedure in the 98SE test machine (AMD Duron 800MHz, 512MB RAM, Amptron/PCChips M810LR motherboard with SiS730 chipset). The error is always the same: Failed to open window station WinSta0 (kexbases\User32\desktop.c:296). As I said in my previous reply, if upon first reboot I replace the kexbases built by me with one built by you or coolman, the rest of my build completes installation and appears to work correctly. So the problem has to be somewhere in the kexbases code and it has to be related to my own developing environment, since you and coolman can build working versions of kexbases. But none of you told me your exact setup so I could check if I'm missing something or need to replace something, such as the 98DDK with the 2000 DDK. The test machine has uSP 3.19 installed, since it initially was prepared for uSP testing. I have no problems building KernelEx 4.5.2, installing and running it on the test machine - the few installed applications behave correctly with it.

Hence my frustration. I grabbed kexbases.dll from your package above, installed my own package built under 98SE, rebooted in Safe Mode, replaced my kexbases with yours and rebooted normally. Everything went fine, verification reported KernelEx installed succesfully. I also noticed there's an addition to the PATH variable in AUTOEXEC.BAT - this is the first time it happens. But then I ran GOM Player. One click on playlist window and it vanished. Another click on main window and that one vanished too. Tray icon was of no use, no context menu, no window show. I had to kill the process using Codestuff Starter; while there, the SizeNS cursor wouldn't show up when resizing the bottom ListView. Then I ran Miranda IM 0.10.11.0 ANSI, built by myself yesterday. Resizing contact list was also funny, no size cursors. At one point, the actual list popped out of the main window and couldn't put it back, mouse wouldn't listen. Then, mouse buttons got reversed: left-click was showing context menu, while right-click was performing single left-click. Couldn't get rid of this behavior even after closing Miranda IM, so I wanted to reboot. Machine wouldn't. I tried from the Start menu, tried from the shortcuts in QuickLaunch (RUNDLL.EXE shell32.dll,SHExitWindowsEx 2 for reboot or 1 for shut down). Error related to RUNDLL32.EXE not being able to run because windows station was shutting down. Finally I managed to reboot by repeatedly performing the three finger salute. After reboot, I disabled KernelEx for GOM Player and Miranda IM and they both behaved correctly afterwards (limited testing for both, though). Oh and the error you get in sdbcreate seems to be caused by your sources being located in 'Documents and Settings' - a path containing spaces which VC6 doesn't quite like. Try moving the sources to a simple path like C:\KernelEx-master or on another partition, anything without spaces or funny characters in the path. Question remains: how come I can't build a working kexbases on two different systems using similar environment?

Now we're getting somewhere! Those functions are still there! All files in the latest zip have the timestamp 2013.03.31 3:27PM (that is most likely the time of the last commit, dunno how timezone weighs in). Anyway, since the fix was supposed to be commited long ago, it's obvious the GitHub was not synchronized at the time. I noticed that by looking at the Release Notes.txt file, which in your builds appears updated with your own changes and version, while the one in the zip is always the same old one by Xeno86. The GitHub also shows that file as being two months old. Since you mentioned Tortoise SVN, I've already stated earlier that I used to have it installed, but removed it long ago because it was heavily slowing down my machine. Besides, the latest 9x-compatible version of it is quite old and - if I'm not mistaken - it may not be compatible with certain newer commands. I vaguely recall there's been a discussion on this topic, but can't recall where exactly. Anyway, I have a feeling your local copy contains some files that - for some unknown reason - have not been synchronized with GitHub. It may well be TortoiseSVN's fault. Please check with the web interface, download the zip and compare files and directories between the zip and your local copy of the sources. Scratch that, I wasn't paying attention. Those functions are not in OpenWindowStationA_new(), but much lower, in SetProcessWindowStation_new(). Sorry for the confusion, I'm too tired.

My VC6 setup on the 98SE machine is years old, I used to compile Miranda IM at the time and I had to set it up properly back then. I'm building all kinds of projects, from Code-Project, Sourceforge or elsewhere. KernelEx 4.5.2 compiles, installs and runs perfectly. Only Leyok's code compiles but hangs at install time. I've posted KEx debug logs, they're all the same. Compilation always flows without a hitch, as I said above I can even compile sdbcreate correctly, so I strongly doubt there's something wrong with my setup. But why am I asking over and over again if you all use the recommended setup and I don't get a clear answer? At this point, the only problem I can think of is you using the Win2000 DDK and not the 98 one, otherwise I really can't see why the same code works for you and not for me. And remember I've said it before: the problem is only in kexbases.dll - if I replace it with one built by Leyok, my whole installation works. I'm gonna try with your kexbases soon, see if it works with it too. EDIT: Just for the sake of it, I reinstalled SP6 on the XP machine, then reapplied the XTREE patch, then recompiled KernelEx. Upon installing, same behavior as before. Here are the logs: link

My bad, wrong assumption about added commands. That CMD.EXE command in coolman's log only shows up when sdbcreate fails to build (as I notice when building under XP), because under my main 98SE I can compile sdbcreate without problems and when it automatically tries to run (twice, as it builds both Release and Debug versions through the full batch command), it only shows a MessageBox saying: "The APPHELP.DLL file is linked to missing export NTDLL.DLL:NtOpenFile." But there is no mention of CMD.EXE anywhere in my 98SE compilation log. Up until this time, I haven't been able to succesfully install one single build of my own, despite everything being built without problems. Neither the ones built under 98SE, nor the ones built under XP-SP3. They all report failing to open WinSta0, install verification never happens (when uninstalling in Safe Mode, there's always two registry keys: "KexNeedsReboot" and "KexVerify" that must be deleted prior to uninstalling). Again, building and installing 4.5.2 under the very same conditions, complete flawlessly and the test system works correctly. Therefore I ask again, for the last time: does anybody build and install correctly from Leyok's code at GitHub, under the following environment? - VisualC 6.0 (Enterprise or Professional) with XTREE patch applied - PlatformSDK february 2003 - Windows98 DDK - NSIS installer 2.44 or 2.46

@ coolman: Haven't checked the latest commits yet, but apparently there's a new pre/post build command that involves cmd.exe, which is not available in Win9x, hence the error. @ Leyok: Yes, I've seen the changes you mentioned, albeit a bit later. I mentioned flexibility in the idea another VxD would ever be added to the project and would require the use of that struct, not only a possible name change for VKRNLEX. But the use of an extra parameter, defined as an array, would take care of the issue anyway, so I guess we cleared this one out. Oh and sorry about the ASM mention - it should've been obvious to me.

As you can see in the EDIT you quoted, Leyok had already taken care of the issue. Generally, I like to keep things open for improvements, but if that's not really necessary for the near future then it may be put to hold. Anyway, this is not my project so I'm gonna stick with compilation issues for now. We'll see what comes.

Well, one thing led to another and this came to attention at some point. We thought "Whadda heck, since we stumbled into it, let's fix it". But it turned out to be harder than expected, at least for an amateur like me. Dunno about Leyok, probably has bigger fish to fry. EDIT: Actually, he commited a fix to this already, although it seems more complicated than it should, considering flexibility is lost anyway. But major problem for me is that none of my builds will work properly. Did you ever manage to compile and install from Leyok's repository succesfully? It compiles fine by me, but always stumbles in kexbases (logs are attached somewhere in previous posts). If I replace mine with Leyok's, it works fine. No idea what's going on, since your code builds and runs fine and I also built and ran many other projects without problems.

Sorry, I tried but obviously my ASM knowledge is way too scarse for this (simple) task. Would you please do the honors? The only other people I know that could cope with ASM are Xeno86 and Tihiy; there used to be PassingBy, but he's long gone and I don't know if Leyok knows any ASM - never mentioned it so far. And just to prove I did try: struct VxD_Desc_Block name##_DDB = \ { \ __asm DD 0 \ __asm DW DDK_VERSION \ __asm DW device_num \ __asm DB ver_major \ __asm DB ver_minor \ __asm DW 0 \ __asm DB[8] name " " \ __asm DD init_order \ __asm DD &name##_CTRL \ __asm DD &V86_proc \ __asm DD &PM_proc \ __asm DD 0 \ __asm DD 0 \ __asm DD ref_data \ __asm DD 0 \ __asm DD 0 \ __asm DD 0 \ __asm DD 'Prev' \ __asm DD SIZE(VxD_Desc_Block) \ __asm DD 'Rsv1' \ __asm DD 'Rsv2' \ __asm DD 'Rsv3' \ } This one obviously works, but loses flexibility: EXTERNC struct VxD_Desc_Block name##_DDB = \ { \ 0, DDK_VERSION, device_num, ver_major, ver_minor, 0, \ {'V', 'K', 'R', 'N', 'L', 'E', 'X', ' '}, init_order, (ULONG)name##_CTRL, (ULONG)V86_proc, (ULONG)PM_proc, \ 0, 0, ref_data, 0, 0, 0, 'Prev', sizeof(struct VxD_Desc_Block), \ 'Rsv1', 'Rsv2', 'Rsv3' \ }

Someone please correct me if I'm wrong but as far as I know, if you're booting directly in DOS mode and not in a DOS window under Windows, anything installed by the SP should not affect the behavior of the DOS applications. DOS games come with their own video/sound drivers and that DOS4GW or whatever it's called, so if anything, it's all hardware-related. Maybe the BIOS on that particular board doesn't do a very good job assigning or separating IRQs and/or other resources. Or you may just have to keep making changes in the BIOS until everything falls in place. Might be time consuming, I know, but as far as the SP is concerned, I strongly doubt there's anything that could help. As for the warning you mentioned, I would guess it's about corrupting the video memory, making the screen garbled and unreadable - something that is not permanent and gets cleared upon reboot. But please, whoever knows better, do chime in!

In newer FF versions (or could it be an add-on?) there are options regarding links behavior when clicked. There should be an option that forces always opening a new tab instead of a new window (FF process). That behavior may be followed by any sites that spawn new windows, but I can't guarantee that. Unfortunately I have so many add-ons installed that I'm not sure which setting pertains to which add-on and my FF is 13.0 so settings may differ greatly. As a starting point though, check out the Tab Utilities add-on, see if it's still compatible (or find and older version) and if its settings allow you to force opening new tab or block unwanted behavior.

What you're experiencing may be an IRQ/DMA sharing conflict. Different video/sound/USB/etc cards use different settings and it's possible that in DOS mode, some of the shared resources create conflicts. You may enter the BIOS if you're confident enough - better note down all settings fo future restoration if anything goes wrong - and try to enable manual assignment of IRQs and DMA channels, then change them around, especially those pertaining to videocard and/or soundcard. If soundcard and/or others are PCI add-ons, you may also swap cards in the PCI slots, because each slot has its own IRQ and DMA and this operation may resolve the conflict (but may as well create another, so be careful). Please note that any such change in the BIOS will be followed by Windows having to redetect devices and in some case ask for the driver, even though it was already installed, so be prepared with the drivers as install kits or already unpacked if zipped. If for some reason Windows malfunctions, put back the cards as they were initially and restore the BIOS settings. Remember: you're doing all this on your own risk. Good luck!

What if we built the struct in Assembler? Would there be a chance to crop whatever name variable to an 8-byte array then? Or define a temporary zero-filled 8-byte array, memcpy the first (max.8) chars (bytes) from the #name variable to the array and use the array in the struct declaration. Dunno if I'm making much sense here - as I said, I'm not a professional programmer.

Indeed, because it'll always append the EOS (NULL) character to the variable string, hence the overflow error. Do you happen to see any viable solution to this, while keeping the flexibility desired by the author? Or should we just leave it as is, waste no more time?

A new browser instance is a duplicate process. It will show up on the Taskbar. It is different than a pop-up, which is a dialog belonging to the current process. Pop-unders are different than pop-ups, otherwise they wouldn't be called differently. In the Exceptions list, try double-clicking the status, see if it changes from Block to Allow or the other way around. (all the above is purely theoretical, I don't have a FF2 installation).