Drugwash

Thank you, it kinda narrows the issue down to 98(SE) vs ME, at least apparently. I really, really wonder if a CJK Windows 9x does experience such issues, because I suspect it's only Unicode vs ANSI here and those versions may be more resiliant to this.

Thanks for the feedback. Do you have KernelEx installed/enabled (for Firefox)? I have the official 4.5.2 installed and running in Default compatibility mode. I wonder what exactly could be the cause of this misbehavior...

Thank you so much for those fonts, loblo! The Internet has become a filthy dump where nothing useful or of quality can be found anymore, not without (heavy) payment or intrusive and shameless advertising. I've been experiencing the same issue as Nomen with Firefox 9.0.1 on 98SE. Unfortunately I can't find the proper settings. If I go to Menu > Options > Content > Fonts & Colors > Advanced > Allow pages to choose their own fonts [...] and enable it, most or all of the text becomes a random series of (capital) letters. if I disable it and go to FontAwesome homepage, for example, the icon fonts won't be displayed, although that particular font is installed. Even with your whole collection of fonts installed, it won't work. Actually it won't work with that setting either disabled or enabled. Default character encoding is set to Unicode (UTF-8), which has an impact on reading/saving some settings files for the current profile and it's best not to tamper with. Does anybody have a working solution for Firefox 9.x on a 9x system...?

Please get to a usable, stable version soon so I can compile and share it. Today I had a fight with an adult ostrich and he won. I'm still alive but next time I may not be that lucky. No, I'm not kidding (this time).

Maybe 256MB of video RAM are too much for this particular hardware configuration. It would explain why it works with lower-end videocards (assumingly with less video RAM). Which means the 6200 card itself may not be faulty but just inappropriate for this system. Also, as I mentioned to Dave-H in another topic, each card may use different memory ranges and IRQs. Try to manually change configurations in Device Manager > Display > ... , if allowed (and reboot afterwards). AGP aperture, video IRQ assignment and other possibly related settings in BIOS should also be revised and tweaked as necessary. Just write down the original settings before you change anything (for both BIOS and Device Manager) in case anything goes wrong, so you can restore them.

There's always a skeleton in the closet. Maybe you should build yourself a secondary machine for a specific purpose, such as video capturing, using perfectly compatible hardware and software, leaving the currently troublesome one to its own tasks with its own compatible hardware and software. Connect the two through a switch or router and you're a happy camper. Unless there are more skeletons in your closet...

Well, you may try to fiddle with whatever settings the BIOS offers - IRQs, memory range etc. - or get a 9x-compatible nVidia card and get rid of the X800 (unless you wanna keep it as spare). Dunno if setting 'Plug and Play OS' in the BIOS (if available) to 'No' would help, but that may create problems in XP/8.1. It's worth a try, just for your peace of mind. There may also be a slight chance of altering the memory and IRQ configuration for the videocard in Device Manager, if the card allows it. I had a Dell machine with a test 98SE installation which wouldn't take an ATI card under any circumstances. I've tried all possible tricks (BIOS, Device Manager) but still yielded conflicts. After pulling the card out and fiddling with memory ranges in Device Manager, the on-board graphics (Intel or something) was accepted and worked fine. So it may just be that your motherboard hates ATI (as much as I do ).

We were looking for the code in the original article, that's why we couldn't find it. I had already found it in the big monthly package but not the separate download. Anyway, the code is the same. The executables in my repository were built from it. I also uploaded the code so anyone can check whether there's any difference. Basically the code may only be useful to someone who wants to try and improve the applications or use portions of it for their own applications.

This is where I'm finding it:https://msdn.microsoft.com/en-us/magazine/cc301756.aspx Yes, that should be the article. For some reason, I'm not seeing any download link for 'Leaks' on that page (should've been at the top, below 'Level of difficulty', if I'm not mistaken). Either one of my Firefox add-ons or some region screening must be playing tricks. WMI is probably required only for the proper operation of the application. I checked the running processes while it was open and indeed WMI was running along. I believe the application assumes WORD-sized process IDs (used in NT- based systems), while 9x uses DWORD IDs, reason why they appear as negative numbers (signed integers) when run under 9x. There may be other bugs/issues too. As for the code at SmartBear: if it were that simple, this would've been a non-issue by now. However, I'll try to write a quick script, see if it yields anything. Don't get your hopes too high!

Well, sometimes mud works wonders against rheumatism... The Norton tool - according to the screenshots provided by Nomen - displays the known active memory blocks only. GDI Usage displays all of the used blocks, included those left behind by leaky applications. The ideal tool would bind both of the above tools' capabilities: display known active handles as well as the stray objects' handles. That and the ability to manually/automatically dispose of the stray objects would make it a jewel for the 9x systems. I'm not sure to what degree Revolutions Pack deals with such objects, if ever.

In XP/8.1 you could leave the default system driver for an older nVidia card and keep it disabled in Device Manager, just as you do with the newer one under 98. If you can afford it it wouldn't hurt trying. I agree with submix8c that the two "giants" don't play well together (usually).

submix8c, that doesn't look like a 9x screenshot, does it? Main issue here is that most tools (except maybe for Norton System Information) don't know how to retrieve and display detailed information on a per-application basis in Win9x. No problems for them in 2000+ though. However, egrabrych needs a 9x report. To confirm, I downloaded and installed v2.0.0.4 of Veign Usage Monitor and the report matches egrabrych's. egrabrych, I had once built a quick and dirty AHK script that could close an open handle such as those reported by GDIUsage. It's not that big of a deal doing that; problem is, closing the wrong handle can screw up the system (current session). That is weird appearance, application/system lock-up, instant reboot... I'm not that good to build an application that could safely close unused handles. But sometime I may play with that thought.

Maybe XP and 8.1 would know how to deal with two nVidia cards. I'd try that first if I had the hardware. Few days ago I had to dismantle an assumingly defective videocard to get the cooling fan for the videocard on this machine, which got destroyed. Funny how I now have PCIe cards that I can't use and miss AGP cards that I so much need.

That's quite odd! SVGACOM works directly with BIOS interrupts and manipulates video memory. It means the driver crash affects communication between the PCIe slot and the chipset and/or other system components. I'm out of my league here, maybe Mr. Loew would know more about this issue. But since you decided to replace the video card, we'd better wait and see how the new one performs. If I was you I would've gotten an nVidia card though, they tend to be more resiliant IMHO. Or maybe I'm just biased due to past experience... Fingers crossed!

I got the same message as above in Firefox 9. Then I enabled HTTPS Everywhere - no dice. I got a principle: if it won't work, I don't need it.

Ok, thank you for elaborating. It seems in both our cases the driver becomes nonfunctional once it crashes and cannot be loaded again. The system probably uses the standard VGA driver instead, until the next reboot. This one is tough. Here's one other thing for you to try, just for the sake of it. Won't fix the crashes or anything, just wanna see if the video memory gets cleaned or not. if it works as I hope, the garble should dissapear. Driver won't be reloaded though, so don't get high hopes. First of all, run Quick Switcher so you can have the hotkeys at hand; you might need to switch a couple resolutions/color depths if the desktop comes back misaligned. Then grab WLL_bundle.7z from my repository, unpack it, go to WLLall > SvgaCom, rename startupOK.bmp to startup.bmp and launch SVGACOM.COM. If everything's right you'll see a fullscreen animated logo (was part of the WLL project started by Tihiy years ago). Hit ESC to end the animation and return to desktop. If the desktop is now clean it means there was rubbish in the video memory, otherwise there may be other issues (possibly bad memory pointers). Use Quick Switcher a couple times if desktop appears shifted to one side. Of course I could be awfully wrong about this all since it's not my expertise. But at least you'll have some fun.

Does my tool return the display to the same garbled screen as per your screenshot or does it show a clean desktop? Just trying to set things as clear as possible. I have my share of display driver crashes with certain applications (most notably Firefox 9 and usually when scrolling down/up a page fast), but I'm using an nVidia card with a very old manufacturer driver that I can't update without losing AGP Texture completely. In my case all bitmaps used by Revolutions Pack for skinning become garbled, everything is slow and the system eventually freezes so I need to reboot as soon as possible. Haven't found a reason, let alone a cure, in years so I'm afraid there's little to nothing I can help here with, except for that tool and maybe some new ideas if they come my way. I don't have Registry Workshop anywhere on my system and neither is it in my CD/DVD software collection, unless it bears a cryptic name. Seeing it's a commercial product, I'm gonna pass on installing it. I also never use Explorer. Ever.

Dunno if it works in such old Firefox versions but it's worth a try: install the FlashGot extension, configure it to enable media download toolbar button (in Options > FlashGot Media tab) and you should see a flashing button in the toolbar when the page contains any videos. Right-click it > Available formats and select a format that suits you (I usually choose MP4). This way you can download the video files and watch them locally with your favorite player (and not to mention keep them stored in case they pull the video at a later time). Personally I keep the Flash permanently disabled through QuickJava, because I do not trust that technology at all and it never worked smoothly for as long as I know it.

In the mean time grab 'Quick switcher' from my repository and set it up on first usage. You get three hotkeys per display where you can set your preferred resolutions. Primary display should work with Ctrl+Shift+F1/F2/F3. When the driver crashes (or changes resolution), use one of the hotkeys to change resolution back to an accepted one. Please let me know if this works as intended.

Appearance is now correct under RP9, disable/enable buttons during processing also works and should be enough of a hint. Thank you very much!

No problems here with Firefox 9.0.1 on 98SE (with KernelEx 4.5.2). Maybe root certificates need to be updated, I know the CloudMe application - which also uses secure connection - fails to connect unless certificates are updated.

Yep, I've already settled that with Dave in PM. Works in XP after modding, waiting for a 98SE test tomorrow. Fingers crossed!

I've succesfully replaced capacitors on motherboards and video cards. Too bad I'm so far away, I could've helped or at least you could've sent me the faulty boards instead of dumping them.

Those poor guys in India or somewhere else, wherever ASUS outsourced their Help & Support, have no idea what they're talking about. There is a monitor driver and it's right on their own web page! But being that they threw everything in there just to get rid of the task... So, to get to the official driver: - open the monitor page here http://www.asus.com/Monitors/VE278Q/ - click Support (top right) - click Driver & Tools in the top bar - in the dropdown list where it says OS select Windows 8 32bit - select Others from the three categories below There you'll find two downloads. First one is ASUS_VE278_Windows_7_WHQL. Click Global below and you got the driver. It has %Chicago% signature, it should work in all 32bit Windows.

I know Bear and it doesn't do what GDIUsage does. It actually lists the elements by handle and displays the loaded bitmaps, brushes, fonts, memory DCs, palettes, pens and regions when selected in the list. This allows the user to recognize the elements and tell whom they belong to. This is gold, compared to Bear. Eventually I found the Leaks package. But it's not in its rightful place anymore. The Wayback Machine used to host a viable page with an active download link, but I guess the "good guys" at M$ got to them - no download link on the archived page copies. See here, for example. There is an alternative location where the package can be found, but after downloading and also searching my 98SE machine, I realized it always was the code, not the executables. Looking in my freeware code folder, I found the three projects - I had compiled them myself on April 9th 2012. Now, since the code has been freely available some time ago at MSDN and I compiled it myself, I guess it wouldn't be a problem to share the executables, would it? If nobody says otherwise, I'll upload them to my cloud and link them here somewhere. EDIT: Uploaded executables and original sources in the Leaks folder (see my signature). Let me know if there's any problem.