jumper

@MiKl: What version of SeaMonkey are you using? See posts #31-33 for SeaMonkey reports. @Joe: I'm not seeing any indication that FF35 calls PrintDlgEx to print. So the presence of ComDlgKs.dll shouldn't make any difference. Please search the FF35 app folder for the text "PrintDlgEx" (else "PrintDlg"). You can also try contents=std,Kstub822,kexbases,kexbasen if "PrintDlgEx" is found. Edit: I just downloaded and extracted Firefox 3.5.19, SeaMonkey 2.6.1, and Opera USB 12.02. Searching for text, I found that all contain "PrintDlg"; none "PrintDlgEx". My ComDlg?? solutions should have no effect on printing in these apps.

Celeron is the budget version of the P2/P3/P4. Socket 370 could be PPGA/FC-PGA/FC-PGA2 pc133 leads me to guess your Celeron is a P2 in a PPGA socket. If the speed is 533Mhz or less, that's it. PPGA runs at 2v. The slowest Socket 370 P3's (Coppermines) are FC-PGA (<=1.75v), so not compatible. Edit: Modifying a PPGA socket 370 motherboard for FCPGA processors

>I've recently installed FF 3.5.20pre.... Initially, any attempt to print would immediately lock-up FF with an hourglass showing. If you had Kstub822 installed, the default definitions would have tried to forward PrintDlgExW to ComDlgKs.dll (from the kstub730 package). If ComDlgKs.dll was missing, the Options message box would have been displayed. Perhaps it popped up under the Firefox window.... There should be a new entry in the log file each time PrintDlgExW is processed--was there one? [note to self: add additional logging of failed forwards in next Kexstubs] >So I downloaded ComDlgEx from post #22, patched this with Import Patcher so that it called ComDlg00.dll instead of ComDlg32.dll and placed this in the System directory as ComDlg32.dll, having renamed the original as ComDlg00.dll. Both ComDlgEx and ComDlgEx2 from post #22 already reference ComDlg00.dll. There are no references to ComDlg32.dll. Perhaps you patched ComDlgEx.dll from post #27 instead? Also, the code for PrintDlgExA and PrintDlgExW in post #22 is obsolete. See posts #27-28 for details. ComDlgEx from post #27 or ComDlgKs from Kstub730 should be used as a starting point for experiments. >Now when I attempt to print, the print dialogue appears (it didn't before), but when I click OK, FF again locks up, this time without an hourglass showing. > >Incidentally, other applications including Opera 12.02 continue to print normally. Also note, the version of KernelEx is 4.5.2 and Kstub 822 is also installed. If you could walk FF3 (with dependencies) through IP35, I'd like to see the .ini file (inside a spoiler would be great). Also, check if printing with Opera 12.02 adds anything to the Kexstubs log file--it may not be using PrintDlgEx. As always, thank you for experimenting and reporting your findings.

Note the CAUTION warning below! Complete directions and board specifications are here: Motherboard Specifications, A8AE-LE (AmberineM) Battery is CR2032 3v, available in the watch department of most drugstores. CR2024 or CR2016 won't last as long, but should work if you use a washer or penny to increase the thickness (as I once did). As a temporary measure, the old battery can be carefully trickle-charged up to ~2v using two weak AAA batteries (as I ... ).

Have you changed the CMOS battery and reset (via jumper or setup) the CMOS data yet?

If you put a copy of the original Comdlg32.dll in the CorelDraw folder, you can customize it (or not) to work best with CorelDraw. The customized version in Windows\System will affect all other apps. If CorelDraw uses a resource template for its extra controls, you may be able to modify that resource to move them down below the expanded listbox.

PrintPDF Firefox extension: What are the symptoms of the printing problems

Use Clipboard Viewer to track what text has been copied to the clipboard. If you select and copy "Rich Text", many applications will not let you paste it. There is a third-party utility that will run in the background, monitor the clipboard, and convert rich text to plain text. Edit: Unicode text could also be the problem. More details here.

>That means that now newer SumatraPDF versions are usable 2.1.1 work great for me. It's the fastest and best of the line and has become my default PDF viewer. > The print dialog is the same like in SumatraPDF 0.8.1. Yes, the visible dialog is still PrintDlg. The data structure is pre- and post- processed to create the PrintDlgEx API. >The dialog appears in SumatraPDF versions 0.9, 1.5, 1.5.1, 1.6, 1.8, 2.1.1. Also 1.9. >I put the COMDLGEX.DLL into the SumatraPDF program folder and "hacked" the file SumatraPDF.exe with HexEdit by replacing the string COMDLG32.DLL by COMDLGEX.DLL (I hope that this was a useful method?). ImportPatcher could have been used, but two bytes are easy enough to change in a hex editor. Well done. Write a better dialog box function instead of reusing PrintDlg. A PrintHookProc could help. Or as a stop-gap measure, I can probably enable the "Selection" radio button: if it's selected upon return of PrintDlg, an additional MessageBox can be displayed that asks if you actually want odds or evens (Yes/No), or the document selection (Cancel).

>That makes the error message go away, but now nothing seems to happen when foobar2000.exe is started.... We can get an app to load by using stubs, but functioning properly is often another story. >Any other ideas? Oldapps.com lists these four builds as being compatible with Win2000. They are our best chance for success. Use compatibility mode "Windows 2000". We don't want to advertise as XP since faking it clearly has not worked! > Has anyone gotten foobar2000 version 1.1.x or higher runs under win-98? >What is shared.dll? There is no Version tab when you look at it's file-properties. A shared dll that plugs-ins use to access the foobar2000 API.

What's in the log file? That will tell us which api's are being used that we can try to do something about. I'm working on a new definitions file that will include stubs to duplicate KernelEx's built-in stubs (but not the better replacements). This will allow us to log all function stubs so we can get a better idea of what is getting called, and when.

Set the compatibility mode for shared.dll to "Windows XP SP2".

IPHLPAPI.DLL (5.1.2600.5512 (xpsp.080413-0852)) does have 'GetAdaptersAddresses !! Maybe this version can be used ? The XP file looks like a can of worms (Pandora's box), but you can certainly try it! I recommend trying Wild Bill's patched version for Win2K, MS11-020, instead. Read the following post as well and get both packages from the links in post #3. In the meantime, I suggest the following definition: Returns an error code (not status) and has five parameters: [iPHLPAPI.DLL] GetAdaptersAddresses=r120s5

The Test 1 results are surprising, but good to know. Test 2 was sure to work using all four options at the same time--and naturally there is nothing in the log file, because the two functions are no longer stubbed! I just realized that the first two options can't help with this problem without addition configuration lines in several sections of Core.ini. However, '3' is a better default than '5' for ME systems and should prevent apps like Dependency Walker from misbehaving. Other apps that do need NT2K mode will need to be individually configured. The SHRegisterValidateTemplate stub should only exist for systems that don't have the IE6 shlwapi.dll. Upgrading the dll is probably better than adding the stub--I'll unstub it in the next release with a comment. If SHRegisterValidateTemplate is only called because GetUserGeoID exists, unstubbing GetUserGeoID may also solve the problem by itself. Otherwise just unstubbing SHRegisterValidateTemplate should also solve things. So after all that diagnosing and testing, it now seems that the solution was to simply look into the log file and unstub the last function handled!

Tests to try: contents=std,Kstub822,kexbases,kexbasen This should work the same as "contents=Kstub822,std,kexbases,kexbasen". My reading of the KernelEx source code is that 'std' is assigned priority slot zero and then up to 255 modules are parsed from "contents=", skipping 'std' which is already assigned. contents=std,kexbases,kexbasen,Kstub822 with these options:Core.ini [ApiConfigurations] default=3 set compatibility mode of Explorer.exe to "Windows Millennium" comment out: [Kernel32.dll] ;GetUserGeoID comment out: [shlwapi.dll] ;SHRegisterValidateTemplate Good research! MSDN says: ERROR_CALL_NOT_IMPLEMENTED 120L is an extended error code for functions that call SetLastError. STATUS_NOT_IMPLEMENTED ((NTSTATUS)0xC0000002L) is a return code for functions that return an NTSTATUS value. We want the extended error code, so these should be correct: [Kernel32.dll] GetSystemWow64DirectoryA=z2e120 GetSystemWow64DirectoryW=z2e120 In the next beta, I'll change the default for 'e' from '1' to '120'

I see no evidence that this is either a DOS app or even a 16-bit app at all. I think it's a Win32 console app. Is there a reason you keep us guessing instead of just telling us the name of the app?

Mpr.dll:WNetRestoreConnectionA exists in Win98se. It is not in my VC5 documentation, so probably does not exist in Win95 nor WinNt 3.51. A few years ago a number of developers reported it missing in this MSDN thread. The answer seems to be Q&A #2 in this Dependency Walker FAQ:

kernel32.dll: HeapQueryInformation is supported by KernelEx with a little help from Kexstubs. Foobar2000 has changed their api and older optional components don't work with the latest app and vice versa. See the FAQ for details.

If your game needs long file names, then it isn't really an MS-DOS app. Perhaps you moved the installed files to a subdirectory that exceeds the MS-DOS path limit. If it really does need long files names, then it must have its own TSR or use a dos-extender to provide that support when running in real MS-DOS. If so, please elaborate. From what I've read on other sites, DOSBox should be the way to go once you've solved the file names/paths issue.

If you have a 64-bit processor, 16-bit instructions need to be emulated. I don't think WOW can do that. Edit: AMD64 and Intel 64 processors can run 16-bit code when operating in "Legacy mode." According to http://en.wikipedia.org/wiki/X86-64#Legacy_mode: Thanks, Tripredacus!

Kstubs816 now posted with better forwarding, better definitions + Forwards can now be abbreviated and commented + Many new forward definitions providing full function support + Complete definitions for all stubbed functions - Stepload support removed Note: Full-line comments got broken last week when I was trying a trick to avoid needing system restarts in some cases. The workaround is to remove from Kstub816.ini / stubs.ini all definitions that are commented out. I've posted a new download with the updated Kstub816.ini. Edit: 822 is what 816 was supposed to be. It has the same features, but without the full-line and end-of-forwards-line comment bugs.

Reflectors: Deep Dish Cylindrical Parabolic Template google.com/images?q=wifi+antenna+reflector Cantenna:

Your antenna is either a VHF or UHF (or a combo) antenna (there is no such thing as an "HDTV" antenna). It is tuned for a few hundred MHz, not 2.4 GHz. The cable also has the wrong impedance for wifi. If it is powered (has an amplifier), it cannot transmit, only receive. I really should just say , but if you're not afraid of frying the wifi adapter (or the antenna if it does have a built-in amplifier) please try it and report your findings! For a variety of reasons, however, you are more likely to have better success with a long wire. For a directional antenna, a "cantenna" is a better place to start than trying to adapt a tv antenna. After much research and experimentation several years ago, I settled on using a seven inch A/B stick antenna inside an aluminum pie tin reflector crudely bent into a pseudo-parabolic shape. My recommendation is to use the biggest wifi antenna you already have along with a metal reflector of some kind.

Already in 98SESP3.3, so should work in ME as well. What app do you run that needs atl80? The only atl.dll on my SE machine is version 2.00.7274 B)

msxml4.dll needs its subsystem version changed to 4.0, otherwise its dependencies look okay for 98se+. It should load as-is with Kex, or after patching. Let us know how the testing goes!