Auto-Patcher For Windows 98se (English)

[soporific]

First, I wonder if 1.92 erroneously skipped the update 'q314941 - Hotfix for Vulnerability in Universal Plug 'n' Play'. 1.95 apparently installs it, even though 1.92 skipped it, claiming I didn't have Universal PnP installed. (At the time, I imagined this might be because I had disabled it in the BIOS at one time, but no, that's no longer true.) I guess now I would suppose W98 has Universal PnP by default, and the update is installed, so I'm good.

More interesting is the IReport.Log shows some weird results. Updates just installed are reported as 'set to by-pass' after the next boot. Maybe this is related to your discussion of reporting bugs. Critical and Recommended is where the trouble starts:

OK, weekend's over --- recovering slightly so please, no loud talking or bright lights ...

Billygoat, you're seeing the results of the big improvements between 1.92 and 1.95 ...

1) AP won't work as intended if there are any 'breaks' in the unattendedness of the entire process. There's no way i want this program to hang waiting for input from the user. Defeats the purpose of the program. So, when you try to install q314941 without UPNP already installed, most of the time you get a prompt saying that "This update was not designed for your version of Windows" (or something similar). Rather than rip the hotfix apart and re-package it (which is what i do these days), i added code to AP to work around this. In fact, the entire DIALOG3 section of MainCode.bat was created to serve this purpose. So, in v1.92 and all versions prior, anytime you saw a message "You can't install X Hotfix, Reason was: Y" meant that the HOTFIX in question would break the unattended status and we couldn't allow that to happen. Most of the re-packaged hotfixes in the hotfixes folder fall into this category. There are probably exceptions to this so don't go jumping to conclusions!

2) and the reporting text has been fully tweaked, those 'Set to by-pass' messages that occur when the previous boot they were saying they were installed was one of the problems that I discovered when testing the Tweaks module (which also had other problems) -- so hopefully the new code will have eliminated all these little bugs.

Submix8c, i'm reading all your stuff don't worry (although you do get a bit cryptic at times!) --- and let me say a big thank you in advance for any suggested code --- it's all good!

EDIT: Please excuse my greediness, but i haven't got any feedback on the debug function -- all good, waste of space, not correct, --- any or all of the above or you could come up with your own words !!

EDIT2: - I've posted the new code in the first post --- hopefully this is the code for the release -- so the good news is -- if you have downloaded the TEST version, all you need to do is overwrite your installation with the attachment. Yay.

Edited March 18, 2007 by soporific

[submix8c]

soporific :

Hope u had a good weekend, OK, DUDE??? (oh, sorry... shhhh!)

1) Thanx for the posted info on the Report "oddity"... it will help understanding the process.

2) Please leave "Debug" in for now... also helps to understand (including Named Variables).

and finally (ta-da-da-dut-da-dah!)...

3) Have solved the "4DOS Installed" problem (yay!).

(3a) AP v1.92 was always (per your suggestion, since it DID "break" otherwise) placed into a ShortPath (eg C:\WINDOWS\TEMP\AutoPach). Using the code I had previously indicated to allow running under 4DOS works in this version (very few and simple changes).

(3b) The new v1.95R2 installer places it into a LongPath (eg C:\Program Files\etc.), hence the code changes to pass new variable "%LOC8%" using quotes to avoid separation at "blanks" during "START/CALL/%comspec%/COMMAND/COMMAND.COM". 4DOS, however (under certain circumstances yet to be determined) expands the variable in the CALL'ed module to INCLUDE the qoutes, so when the next CALL to a module is attempted, the instruction DOUBLES (shh!) the number of quotes, and causes it to error out.

(3c) The second aspect of 4DOS is the way it handles several "ASET" instructions; the literal string (anywhere within a string!) %+ is reserved in 4DOS, hence the error.

(3d) The third aspect of 4DOS is that any and all errors will always be displayed, and any instruction not containing " >nul" will be displayed. This does not actually cause a problem, but isolates those instructions where this method wasn't (it was elsewhere...).

I have completed a VBScript for v1.95R2 (and above) that changes the LongPath to a ShortPath and places the value(s) into a file, to be subsequently read to SET "%LOC8" to the ShortPath to "beat" problem (3b). It works fine (about a dozen lines of code) as long as the extra (new) quotes are removed where necessary within the appropriate ".bat" modules.

Problem (3c) only occurs in v1.95R2 and appears to only be in the reporting/date conversion routines (%numvar%+1) in "AutoPach.bat". This can be fixed by using the following method (per ASET documentation): wrap both values with single-quotes, as in '%numvar%'+'1', which eliminates 4DOS parsing.

"Problem" (3d) above, as I said, can be completely ignored or corrected as you will. I intend to "fix" a copy for my own purposes (the TABLES method...) so I can utilize the 4DOS functions while including 4DOS in the package as a prerequisite. Which reminds me, why are the @Regtest.txt file(s) so large? Under FAT16, it indicates that 3 (count 'em - 3x32kb) blocks are being used in may cases. What on earth are you writing to it? Or is this a "my bad, must be another 4DOS idiosyncracy" deal?

It works! However, there appear to be some generated "text" files that are being processed (in one routine only!) that appear as "Not Found". Don't know what instruction was used against it/them, but will isolate and correct, since I am assuming it was pre-existing even for non-4DOS; 4DOS just helped identify it.

Will get back to you after I have completed only the necessary (so they will be minimal) changes to support bi-%comspec%-compatability. Which reminds me (again); in some places, you use %comspec%, in other places you use COMMAND.COM, and in one single place you use COMMAND (???).

And finally, where did 3dParty Components go? Is this intended to be in a separate/new package along with many other wonderful Freeware utilities (download managers, etc.)?

---- billygoat:

REPEAT! If you installed that particular Optional Component at ANYTIME WHATSOVER, you DO have UPnP and it cannot be uninstalled (I think...)! This is part of the XP-OS and can be obtained even without AP! Search the inet for it and do some light reading; it will clarify you understanding of it's purpose.

Try another clean install and be selective when patching (i.e. select ALL, then go de-select the indicated Optional Component). If you don't have the time or resources, someone else may do it for you. It could be a possible "bug" (hope not!).

----

(One last note: the Debug file can get LARGE, and requires WordPad to open; better install it, guys!)

EDIT: OH S***! Apologies... All references above to v1.92 are incorrect; actually v1.7 DOH! Will quick-check v1.92 to confirm; otherwise v1.92 and v1.95R2 will be considered to be in the same pickle...

WHEWWWW (shhhhh)!

Edited March 18, 2007 by submix8c

[nathanson1947]

Soporific, et al:

DID have a problem with WU (using the Start Menu/Windows Update), and went to folder WINDOWS, created a shortcut to WUPDMGR.EXE and placed it on the Desktop. The difference in the two shortcuts is the "generated" one has "Start In C:\WINDOWS". THAT one worked (beats me )!

This seems to work with the WUPDMGR.EXE at the following location on my computer:

C:\WINDOWS\VCM

[Drugwash]

This is getting a bit too complicated, with VB scripts and all. DOS batch is OK: I can read it, edit it if needed...

Just a personal thought.

[submix8c]

Drugwash :

Don't worry about the VBScript suggestion. It's mainly for my own purposes. Forget the 4DOS support. It's for my own purposes. Read my previous posts. I had mainly intended to use it to "revamp" AP into a CDRom-style install/execution. If it works/does-not-work out that way (incorporated into AP), then so be it. I'll be satisfied that it works. I'm already near completion of 4DOS support (Phase 1). Phase 2 is utilizing 4DOS by modifying current processes to use tables and generic processing to shrink the codeset and to allow for easier future modification, mostly via the tables not the codeset (need a new 3dParty/Optional? just plug its definition into the table! menus and installations are DYNAMIC!). Phase 3 is to modify so execution is from CDRom instead of from HDD. Again, my own needs. Just thought it may be useful to the Project.

FWIW, here is the VBScript source:

'ShortPath.vbs - Gets Short (8.3-style) Path Name.
'© None... needed to get it, so I did.
' from within a ".bat" file
'        start /wait "%LongPath%\shortpath.vbs" "%LongPath%"
' one-line ShortPath result is in %LongPath%\ShortPath.txt
Option Explicit:
Dim sp, fso, f, spcf, sptf
' Get the passed variable %LongPath%
sp = WScript.Arguments(0)
Set fso = CreateObject("Scripting.FileSystemObject")
If sp = Empty Then
Else
 Set f = fso.GetFolder(sp)
 sp = f.ShortPath
End If
' Write a one-line file containing ShortPath into LongPath\ShortPath.txt 
Set fso = CreateObject("Scripting.FileSystemObject")
sptf = sp & "\ShortPath.txt"
Set spcf = fso.CreateTextFile(sptf, True)
spcf.WriteLine(sp)
spcf.Close

I perform this after obtaining LongPath (the current way), then "read" (using NSET) the one-line ShortPath into another variable.

VBScript is, as I had said about DOS Batch, a Scripting language, as is Basic, 4DOS, JScript, Perl, and many, many others. The programs used in this project are also "3dParty" and have specialized functionality (fairly well docomented), similar to non-DOS-batch Scripting language Functions (bet they are written in old Basic, Assembler, or Borland C/Pascal!). Learn all of them you can. It makes you a better programmer. I'm a little gray around the edges but still figured out how to do the above VBScript... (p.s. notice the "Change User Name and Org" component is a VBScript?)

Don't be alarmed; all is well !

Edited March 19, 2007 by submix8c

[billygoat]

I just finally got a chance to try the new code, but haven't looked at the release link yet. The AutoPatcher 1.95 installer I used was still putting kb931836 - Unofficial 2007 Updates to Time Zone Daylight Saving Time - in Manual Updates, but the Critical and Recommended Updates module now expects kb931836.exe in Hotfixes folder. AutoPatcher complained about this during install and notes the missing update file in the log. (I also see this wasn't the same executable used in 1.92.)

I've not had a chance to test some of the modules yet, either, but hopefully others have done so. I guess Tweaks is good to go now, but I didn't use it.

I've only (so far) used:

Stability Hotfixes

Windows Installer

Internet Explorer

DirectX 9

Critical Security Hotfixes

Critical & Recommended System Updates

ASPI Layer

Win98 to Me

(The worst thing about 'Win98 to Me' failing on my system was it was the last update, FWIW...)

I see the wait time has been reduced. Reminds me, I noticed after the DirectX installation there was still lots of hard drive activity when the timer ran out, even when it was set to 20 seconds previously.

Log looks correct now though. :-) (No more 'set to bypass' anomalies...)

[soporific]

I just finally got a chance to try the new code, but haven't looked at the release link yet. The AutoPatcher 1.95 installer I used was still putting kb931836 - Unofficial 2007 Updates to Time Zone Daylight Saving Time - in Manual Updates, but the Critical and Recommended Updates module now expects kb931836.exe in Hotfixes folder. AutoPatcher complained about this during install and notes the missing update file in the log. (I also see this wasn't the same executable used in 1.92.)

Log looks correct now though. :-) (No more 'set to bypass' anomalies...)

A good side-effect of a 5 day wait to upload this new version is that i went over the program with a fine tooth comb and made little changes here and there. The 20 to 15 second delay after reboot was one of these. Regarding disk activity still going after the DX9 installation, and after thr 15 second delay --- this shouldn't be a problem --- if there was an immediate reboot, maybe, but no module does that - whatever was still loading when AP restarts should be fine. Yes, kb931836 should be back inside its module --- that's why you'll need the release version (looks like i was wrong about only needing the code to upgrade from TEST to release.) -- yep, reporting text should all be good -- i did quite a bit of teaking --

Did Win98-to-Me fail again? Let me know ...

Edited March 20, 2007 by soporific

[Drugwash]

Just some quick notes. I downloaded and installed 1.95PR2, then I extracted the newer files from the zip attached in first post over it - dunno if that was a good or bad idea. I chose a lot of modules and options to install, but I couldn't say which - the log is on another machine and there's no way I can transfer or publish it here, currently - I'll try attaching a floppy or something.

Thing is, there were three issues that broke the unattended operation:

• while installing Dial-Up Networking 1.4 Upgrade, the system asked for the Win98SE CD. The files were already copied in some folder on the HDD, but currently there's no (or bad) code to check for the files in other path than the original install path.

• at the end of some modules' installation, when reboot should've occured, warnings popped up instead, saying something like "You must quit this program first if you want to quit windows. Press Yes to close this program and quit Windows, or press No to continue running this program and Windows". Choosing either option would reboot, but the dialog popping up and requiring an action was a show stopper, and it did pop up quite a lot of times (3 or 4 times, at least).

• many times, the machine wouldn't reboot - the "Windows is shutting down" screen would stay on for minutes, keyboard would not respond; only way to "end it's pain" was pushing RESET. This may be a BIOS/Hardware issue or not, as generally, the reboot went fine. The shutdown/hang hotfixes have been applied, but it still behaved after that.

Other than that, I grew old. Almost three hours of installing and (assisted) rebooting. Oh, well... I still have work to do, with manual installs and then check the log - didn't have time for that last night, and I'm gonna be on the road these days, apparently. Will check back in later, time allowing.

[billygoat]

Did Win98-to-Me fail again? Let me know ...

Okay, I got the release version and tried again with that same restored clean install C: - same result on the Win98-to-ME module installed by itself. (If I had time, I'd go ahead and try MDGx's original, but there's other priorities.)

I installed the same options as before, everything seems fine. Still to do: Tweaks, WMP and Other.

Oh, BTW, guess I've missed info on the System.ini stuff.

Followup on changes to the UPnP vulnerability Fix: If I understand you correctly, the repackaging you've done now allows the install even when UPnP isn't installed. (The original package would abort.) And further, I'm guessing, if I elect to install the XP Networking tool in the future (no desire right now) it would be wise to only use AutoPatcher to do that, so as to insure the fix is still in place.

Thanks again!

[ENDer2000]

Working download links:

http://www.softpedia.com/progDownload/Auto...load-64269.html

10x to soporific

Edited March 21, 2007 by ENDer2000

[billygoat]

· at the end of some modules' installation, when reboot should've occured, warnings popped up instead, saying something like "You must quit this program first if you want to quit windows. Press Yes to close this program and quit Windows, or press No to continue running this program and Windows".

I had this happen after the MP9 module today, but I also had this happen once before with an earlier version of AP (different module, I think) without being able to replicate it, so I didn't report it, thinking it was peculiar to my system or a random occurrence.

[soporific]

· at the end of some modules' installation, when reboot should've occured, warnings popped up instead, saying something like "You must quit this program first if you want to quit windows. Press Yes to close this program and quit Windows, or press No to continue running this program and Windows".

I had this happen after the MP9 module today, but I also had this happen once before with an earlier version of AP (different module, I think) without being able to replicate it, so I didn't report it, thinking it was peculiar to my system or a random occurrence.

The message you are getting happens when AP has lost control of the OS during a patch session -- by using the switch /Q:A /R:N with most hotfixes, what we are doing is asking the hotfixes, nicely, if they can leave any rebooting up to the user. Unless its an actual bug (still possible), at least one update is rebooting after its finished when its not supposed to.

If you can narrow down which ones are doing this and post the result, that would be helpful.

[Max_04]

@ soporific:

I'm ready to realize AP for italian system, with a reduced number of components for win98se (28).

Therefore I'm localizing AP 1.95 code.

For you it is possibile to automate execution of IEradicator and successive installation of IE6SP1?

If yes, there is no problem.

Ps: I send you also all components interested in a 7zip archive more later.

Edited March 21, 2007 by Max_04

[MDGx]

soporific:

If I may make a suggestion:

If you'd like to add more options/system components/updates/etc to Auto-Patcher, you could add only the direct download URL links to the actual files [eventually with a short description], and let the user select which ones he/she wants to install.

This way new add-ons won't increase the size of Auto-Patcher.

HTH

[Drugwash]

No. There are still many people that do not have Internet access. Auto-Patcher can be copied over from a CD/DVD media and installed on such system, but relying on download links would just cripple the package.

That could only be a temporary solution for alpha/debug/testing versions, but definitely NOT for release versions.