jumper

I program 100% on 98se. KernelEx4.52 is installed and defaults to compatibility mode Windows 98 SE (for the stability enhancements). Incompatible: CrystalCPUID (for over/under-clocking my C3 cpu) and KernelEx (all settings). (6) The Kext topic includes many success stories for KernelEx+Kexstubs. (Browse above and below for many more.)

>Though not as fast as RAM memory it is faster than accesing the harddrive I'd rephrase this as: "Though not as slow as some harddrives, it is far slower than RAM." Since I won't be working on a new solution anytime soon, let's look at existing options: 1) Add as much RAM as possible 2) Add more and faster disk drives: + faster: newer/larger/SSD + more: one each for system, swap, data 3) Put system files on thumb drive, ConservativeSwapfileUsage=1, minfilecache=1024 4) Put swap on thumb drive, ConservativeSwapfileUsage=0 5) Use compressed FAT-16 on thumb drives to speed transfers 6) Use CF with IDE adapter as poor-man's SSD

Shell32vlc.7z Shell32vlc.dll is a wrapper that forwards three calls as needed (by VLC 1.1.10): LIBRARY Shell32vlcEXPORTS CommandLineToArgvW = SHELL32.CommandLineToArgvW SHGetFolderPathW = SHFOLDER.SHGetFolderPathW ShellExecuteW = SHELL32.ShellExecuteWPlace in app folder and use in one of two ways:1) rename to Shell32.vlc and use ImportPatcher or a hex editor to change PE references from Shell32.dll to Shell32.vlc in files that need SHGetFolderPathW 2) rename to Shell32.dll and use RegEdit to delete or rename the SHELL32 entry in KnownDLLs (HKEY_LOCAL_MACHINE\System\CurrentControlSet\Control\SessionManager\KnownDLLs) Edit: Method #2 won't work with this version of the wrapper because Shell32.dll can't forward to Shell32.dll!

Yes, I have seriously "thought of making ready boost for Win9x". No there isn't one already. It's a big topic. Please describe your system and what are you hoping "ready boost" might do for it.

I just tested in Firefox 2.0.0.20 on 98se and "Add to Post" worked as expected. Make sure you have any ad/element/script blocking disabled. Also check the Error Console for scripting errors. Remember to include software and OS versions when reporting problems.

Plug-ins are usually just scripts that reuse the code that is already in Firefox. If FF3 can't print, I wouldn't expect a plug-in to be able to, either. Have you tried version 0.59?

Open "My Computer" (or any Explorer window) and enter %tmp% in the address field (also try %temp%). You can also use Find (Files and Folders) to search your disk for files modified in the last day. If your browser is wrapping the IE engine, perhaps the default Bookmarks and History methods just save to memory. If so, you will need to read them from memory and write them to disk upon app shutdown.

A year or so ago, Youtube updated their video interface and the controls disappeared for all versions of Flash 9.0.*. Earlier today I tested over a dozen Flash versions (from 9.0.47 through 11.1.102.63) and confirmed that to see the controls, you will need Flash 10.0 or later. That means no more Youtube video controls in NT4.

When you test the app, does it run interpreted inside the IDE/sandbox or are you compiling it and running it stand-alone? While the app is running and you have "saved" some bookmarks, look for any (temp) files that are created or modified on disk. Temp files are normally deleted automatically when closed or the app ends. If there are no files, then the bookmarks and history are only being stored in memory and you will need to find a method to save them to flat files or a database.

I've just downloaded the MPlayer source for these two versions and will look for changes that might have caused the page fault. Update: In r36568, 101 (of 5870) files were changed. In "Changelog" this line looks like the likely culprit: * Updated build environment to GCC 4.8.2 and MinGW-w64 v3.

Flash by OS: NT4: Without SSE, Flash versions later than 9.0.47 are out of the question for NT4. 98 : For Win98, however, you can use FineSSE to overcome the SSE issues of later Flash 9 versions (up through r289). 98+KernelEx: Flash 10.0-10.3. Opera by OS:NT4: 10.63 is possible; 98 : 10.10, 10.63? 98+KernelEx: 10.63, 11.11, 11.64, 12.02 Firefox by OS:NT4: 1.5.x, 2.0.0.20, 2.0.0.22pre 98 : 1.5.x, 2.0.0.20, 2.0.0.22pre 98+KernelEx: 3.0, 3.5, 3.6, ... For NT4, sdfox7 recommends Opera 9.64+Flash9.0.47+SP6 (not 6a) (see post #2). See also this update (#4). For Win98, I recommend Firefox 2.0.0.20+Flash 9.0.47 (I use this daily). For 98+KernelEx, I suggest trying Firefox 3.6.x and Flash 10.1; Opera 11.64 is another option.

For Flash 9 after version 47, cpu becomes an issue. SSE is required for video and that means a PIII or greater. What is your hardware setup?

After reviewing post #1, I'm confused. 1) What OS are you using? 2) What browser are you using? 3) What is your goal? 4) After two weeks, why are we covering the same ground in yet another thread? "I also used the modified Flash file for Opera 10.63, and now Youtube runs flawlessly in Opera 10.63."

1) Failing to load is _not_ "a crash" 2) 9r47 doesn't call MonitorFromWindow, 9r115 and up (including 10.2 and 10.3) do. 3) 9r115 and up also require SSE for videos 3) 9r47 is very stable in FF2..20 on SE, but not so in Opera 10.63 You asked about getting 9r47 to work in Opera 10.63 on Nt4. Please describe in detail exactly how it fails. Your best option may be to hope that 9r47 is stable in Opera7 or 8--whatever is the earliest that can display a minimal Youtube page or embedded video. Also confirm that you have NT4sp6 and not 6a.

If scanregw.exe has been running automatically at startup, you should have a registry backup from each of the last five days. Boot to DOS and "scanreg /restore" a backup from before things went wrong. If no registry backup is available, change the IP address back to an *invalid* value. I suspect the black screen is caused by a (LEGO server) app that runs at startup. Without a network connection, it aborts to the desktop. With a valid connection, it waits for a remote (robot) app to talk to it. The black screen is probably a background for displaying webcam feeds on top of. Pressing the power button is causing Windows to shutdown normally. Instead of pressing the power button, you can try Ctrl+Alt+Del to bring up the task list--kill apps until you get a desktop (or have to shutdown the system). Once you get to the desktop (by safe mode or other method), run msconfig.exe and use the startup tab to disable extraneous (LEGO) apps from running automatically at boot. Then it should be safe to reconnect to the network.

technoid, If any of the FSB tweaking tools you mentioned supports your mobo/clock chip, then boot at 83 with a divisor of 3 and then up the FSB to 90 for a PCI bus at 30 and CPU at 540. If that works and you have a good heatsink, maybe you can overclock: * FSB : 95 , PCI: 31.6, CPU: 570 * FSB: 100, PCI: 33.3, CPU: 600 Another possibility is to stabilize the PCI bus at 41.5 by disabling some of the PCI speed-up features (PCI burst, etc.) in BIOS.

If the BIOS sees the flash drive as a fixed HDD, it may be changing the normal drive letter assignments. Try disabling real-mode USB support in the BIOS.

What is the history of this drive?

In Win98, MORE doesn't convert the EOL, but SORT does. Using the pipe as the temporary file, this works (without sorting!): type %1 | sort /+-1 > %1

Or once the source folder is empty, the last block move can be undone--but that can also easily be done manually.

Sorry I've fallen so far behind in responding, but here are some new discoveries: The other day I realized I could monitor usage of existing standard functions by having Kexstubs log them and then pass the call through: [User32.dll] MessageBoxA=>User32: ;log and pass through Log entry each time a MessageBox is created:= User32.dll:MessageBoxA=>User32: ;log and pass through = This led to another discovery: Logging can be enabled/disabled on-the-fly with just an app restart (just like changing a definition).Here are scripts that will quickly set the logging mode: Kexstubs Log=0.vbs ------------------ set o=CreateObject("Scripting.FileSystemObject") s=o.OpenTextFile("Kexstubs.ini",1).ReadAll o.OpenTextFile("Kexstubs.ini",2).Write(Replace(s,"Log=1","Log=0")) wscript.echo "Kexstubs logging is OFF" Kexstubs Log=1.vbs ------------------ set o=CreateObject("Scripting.FileSystemObject") s=o.OpenTextFile("Kexstubs.ini",1).ReadAll o.OpenTextFile("Kexstubs.ini",2).Write(Replace(s,"Log=0","Log=1")) wscript.echo "Kexstubs logging is ON" Kstub822 Log=0.vbs ------------------ set o=CreateObject("Scripting.FileSystemObject") s=o.OpenTextFile("Kstub822.ini",1).ReadAll o.OpenTextFile("Kstub822.ini",2).Write(Replace(s,"Log=1","Log=0")) wscript.echo "Kstub822 logging is OFF" Kstub822 Log=1.vbs ------------------ set o=CreateObject("Scripting.FileSystemObject") s=o.OpenTextFile("Kstub822.ini",1).ReadAll o.OpenTextFile("Kstub822.ini",2).Write(Replace(s,"Log=0","Log=1")) wscript.echo "Kstub822 logging is ON" Kexstubs Log=n.7zEdit: To monitor standard functions, Kexstubs needs to have a higher priority than them. So in CORE.INI: contents=std,Kexstubs,kexbases,kexbasen - or - contents=std,Kstub822,kexbases,kexbasen Originally edited December 21, 2013 by jumper

Scrolling all the way down to the bottom, I still see no replies by other users--just the one by sdfox7 himself. If you mean one of the "Related Solutions", please give that title and link. (Searching both the page and the page source for "Active Desktop" gets no hits.) See two updates at: Update: YouTube Functionality on NT 4.0 Windows NT 4.0 Service Pack 6 is recommend; Post-Windows NT 4.0 Service Pack 6a Security Rollup Package is known to cause problems. "GdiGradientFill could not be located in GDI32.dll" error on startup Re: GradientFill vs GdiGradientFill (GDI32) NT4 does not have GdiGradientFill. If you install (anything that installs) a W2K+ version of msimg32.dll, it will look for GdiGradientFill and fail. Uxtheme.dll is another likely culprit.

How about an HTML/webpage editor? Ideally a word processor that can save your formatted text as a webpage and then import transparent graphics that can float over the text and be positioned as desired. The result could be saved and viewed in any web browser. The text would remain searchable, copyable, editable. If words and paragraphs are <span>ed, scribbles could be positioned relative to any one of them and stay with them as they reflow.

FF3 is never mentioned. It says FF2.0.0.20. And the registry mod was for Flash8, not FF. What is the text of the GDI32 error message? Try running Dependency Walker or ImportPatcher on firefox.exe to see what API's NT4 might be missing.

Dependency Walker 2.2.6000 is very good, but must be used to test on the target system. It can also profile an app after initial testing to determine whether there are any additional landmines (but this would fall under "run it in 9x physically"). >...(although it doesn't say which file version is required)... This feature might be worthy of a project--I don't think there are any tools that can do this. There are clues in the PE file headers and other structures about when each module was linked (created), for what target platform, by what linker (for crt version info), and possibly to the timestamps of other modules (if bound). While most PE tools will indirectly provide most of this info, I don't know of any that summarize it in terms of a target system. And ImportPatcher is the only tool I know of that reports binding timestamps (valuable for this purpose).