Recapping, build 142 only runs in a Windows environment, right?

So is it possible to run this build on a virtualized Windows environment like VirtualBox, VMWare Player and possibly QEMU? I'm not exactly sure as to how QEMU virtualizes things!

I ask because if I can get something like LIMBO to run Windows 95 or higher on my Android environment, I could try that special build on my Chromebook!

I want to both thank you Lauri for all you've done to keep 68k Macs in the spotlight. Without all you've done, a huge part of the world of those first beautiful Macs would have been lost to time!

I can't wait to hear your life/Mac stories and anecdotes! In the covid-zoom world we're in now, maybe a zoom-youtube interview special that can later be ported to the cornica.org site would be an amazing honor and treat to everyone here at the Garden, as well as those who love and use Basilisk II daily, like myself and many, many others, and far beyond all of us!

I hope and pray your health improves, Lauri!

One thought that came to me as you talked about Windows compatibility is the possibility of using a virtualizable surrogate OS.. I was thinking about ReactOS, or XP as Mike mentioned. I, like you, absolutely loved the introduction of Windows 7: it was so balanced, uncluttered and it's a shame time has taken this approach to OS simplicity away compared to what we have now in Windows 10. I meantioned ReactOS because I like how it keeps everything very simple and low-specked, and that you can virtualize it to run it on any modern OS, including Linux!

I know, I know, it's Alpha software that's been worked on for the past decade-and-a-half, but it may work for running Basilisk II anywhere you need to, just a thought!

Thanks again for joining the Garden, Lauri!

Apparently it is ok that I tell stories here. I will continue for a while, at least until someone says "shut up"

The story of "cdenable" is worth telling.

Back then, many developers noticed a very strange and annoying problem: under Windows 95/98 series, it was not possible to read CD sectors at the beginning of the cd, the so-called ISO superblock. Microsoft had put in some limitations to prevent it. I noticed it when I was working with HFVExplorer. I started HFVE as a learning project, I needed to know how to use MFC (Microsoft Foundation Class library) at work, and this was a perfect opportunity for me to study. I will write the story of HFVE later, it is interesting and requires its own page.

However, I could not let that be a roadblock. Hah: when I was younger, my mother said a couple of times that she had never met a person as stubborn as me -- and trust me, that was not a compliment!

Ok. I launched SoftICE -- the kernel debugger I used back then -- and followed the code path downwards. Down we go, down down ... like Alice in Wonderland, down the rabbit-hole ... I noticed how I stepped into ring 0, kernel mode now ... and there it was: a purposely put in denial, no reading of the first 0x10 sectors! Why? I could not understand the reason for this, and it is still a mystery to me. Some Microsoft - Apple commercial warfare?

The only way to circumvent this was to write a vxd. And oh boy, this was a messy business! I had never done this before, and the code samples that Microsoft provided were horrible. I wrote one anyway: it patched the memory image of the cd driver interface on the fly to remove this stupid limitation.

"cdenable.vxd" was not the most stable piece of software ever written. I received many bug reports. Some of them I could fix, some remained mystery. When APM-aware devices started to emerge more, the driver needed some rewriting ... all in all, many times I regretted that I wrote it at all! Because I didn't even use 95/98 series of operating systems normally, I was an early adopter of NT.

But regardless of those headaches, in 1997, something happened that made cdenable worthwhile: I was using Executor, and liked it very much. Because at that time, there was not yet anything better. I wrote a couple of utility programs that linked to Executor. What did they do? I cannot remember. Launchpads, maybe? I uploaded them. And had second thoughts. Is this okay? I wrote an email to Clifford Matthews, the man behind Executor.

He was a friendly man, a scholar, polite, humorous, and had some health issues. He wrote back to me a standard reply, very politely: yep yep yep, do what you want, no conflicts

Then, 3-4 days later, came another email. He had realized that cdenable.vxd existed. Executor had the same problem as everybody, reading the first sectors, and that was essential for a Mac emulator. I gladly gave him permission to use my driver and it was included in many future Executor releases. We talked extensively for some years. I always helped him when I could, he was a very charming person! And oh yes, I remember being so proud when Cliff included my name in the About box credits of Executor! Haha!

I also contributed slightly to the Windows version of vMac. Did it use cdenable too? I think so.

There was a DOS Mac emulator made by Microcode Solutions, "Fusion". Interestingly, after cdenable had done its magic, that is, patching the memory image of the cd driver, Fusion could read Mac cd's too! A nice side-effect. This was needed only when Fusion was used in a DOS-box of Windows; when the computer was booted in pure DOS mode, it had no problems reading cd's.

Writing cdenable version for NT series of OS was no problem, easy-peasy for a professional programmer. There were no such strange roadblocks, by writing a standard .sys driver all CD sectors could be read. I don't remember receiving any bug reports regarding that one. I don't know how things are nowadays, perhaps no driver is needed any more ... at least that would make sense.

And think about how easy stuff like this is under Unix/Linux ... cd's are just block devices. You just read!

Lauri

Series of interesting email exchanges, nro 0001:

Haha

Lauri

UNHAPPY "CUSTOMER":

Due to your software, my hard drive has suffered file system damage. I created a volume of 500MB, Using your software. Originally my Hard Drive had 1.3GB free space. While transferring from volume to volume (30MB worth of files from a 40MB volume, to the 500MB volume, your software crashed.. Leaving my hard drive with 7.5MB free space. This is NOT RIGHT!

I want to know
1.) Where has the disk space gone
2.) Why does your software do this
3.) How can I get the disk space back.

If there is permanent damage to my hard drive, I will make sure this potentially lethal software is removed from the internet. Why should 1.3GB be used up to transfer a 30MB file?

SUPPORT GUY LAURI:

>Due to your software, my hard drive has suffered file system damage. I
>created a volume of 500MB, Using your software. Originally my Hard Drive had
>1.3GB free space. While transferring from volume to volume (30MB worth of
>files from a 40MB volume, to the 500MB volume, your software crashed..
>Leaving my hard drive with 7.5MB free space. This is NOT RIGHT!

I'm assuming that you are talking about HFV Explorer.

>I want to know
>1.) Where has the disk space gone
>2.) Why does your software do this
>3.) How can I get the disk space back.

Sounds like you are using the new Windows Millenium Edition which has a silly "System Restore" feature, effectively taking backups of all files, all the time. Including HFV volume files.

>If there is permanent damage to my hard drive,

I don't know any software that could actually damage hardware.

>I will make sure this
>potentially lethal software is removed from the internet.

Huh?

>Why should 1.3GB be
>used up to transfer a 30MB file?

Turn off Windows ME System Restore, or whatever it is called, and delete the garbage backup files.

And DO NOT email me ever again. I have zero tolerance for threats.

"CUSTOMER":
I am very sorry about the previous mail I sent. I was very stressed out
because things like this have happenned before and has actually destroyed my
Motherboard, CPU, Hard Drive and RAM, Luckily my PC was still under warranty.
The hard drive had lost so much space that it had overloaded and literally
made the motherboard burn out.

I didn't mean any of the things I said, I'm known for having a filthy temper.

It seems running the software again has given me 1.3GB, But system restore
has still taken most of my hard drive space. It hasn't made any restore
points before the software was run, So I will complain to micorosft..
unlikely that they will actually help though.

Once again I am very, very sorry.

Series of interesting email exchanges, nro 0002:

The famous (infamous?) HFVExplorer lawsuit! Who remembers this, rise up your hand!

HFVExplorer has been included in countless CD's, those that ship with magazines. Some of those publishers asked for my permissions -- which was actually unnecessary -- and some didn't.

All was well until in Germany, a company named Symicron sued both Microsoft and the famous c't magazine in court -- why? Because Symicron had trademarked the term "Explorer", but never used it in any of their products! They sued Microsoft because M$ had "Explorer" file manager and "Internet Explorer", c't because they had distributed two programs on their add-on cd's, HFVE and FTP Explorer.

I remember that I was quite irritated ...

All the information available regarding this was in German language which I never mastered. Christian translated it for me, I let him tell you the rest of the story. I'm sure that he doesn't mind, I checked that there is nothing personal included. These emails are dated 2002:

("DM" means Deutsche Mark, the currency they used before Euro in Germany. In Finland, we also used marks -- Finnish Markkas! Haha)

Hi!

On Sun, Sep 22, 2002 at 11:22:45AM +0300, Lauri Pesonen wrote:
> >but some years ago the German magazine c't was sued over the issue
> >by Symicron. They claimed to have exclusive rights
> >to use the name "Explorer" or something to that extent.

According to c't (http://www.heise.de/newsticker/data/hob-30.07.02-001/),
Symicron's "Explorer" trademark has been deleted. The first paragraph reads:

  Today, the German Patent and Trademark Office made a board decision to
  delete the German trademark "Explorer". This trademark was previously owned
  by the company Symicron based in Ratingen. The reason given for the
  deletion in the statement available to heise online reads: "The department
  of trademarks is convinced that the owner of the trademark in question was
  in bad faith at the time of the trademark's registration on September 22,
  1995 [hey, that's precisely 7 years ago :-]. The specific acts of bad faith
  that come into consideration are the matters of obtaining a trademark by
  [fraud/trickery?], of a "blocking trademark", and of an "ambush trademark"
  [I'm not certain what this is supposed to mean; maybe something similar to
  a "submarine patent"?].

However, the last paragraph says:

  Currently, the decision has not yet become final. It can still be appealed
  against at the Federal Patent Court within the period of one month.
  Symicron's lawyer von Gravenreuth already announced that he will recommend
  this step to his client.

Well, this "one month" is over. I don't know what the final decision was,
though.

Bye,
Christian

Hi!

On Mon, Sep 23, 2002 at 02:10:12AM +0300, Lauri Pesonen wrote:
> Sounds like Symicron was on a fishing trip just like the people
> registering domain names with the goal to sell the domain
> to a company which only later on realizes that they have
> an interest to the given name -- maybe that's what an "ambush
> trademark" refers to.

Yes, that's the impression I get. Apparently, they never used "Explorer"
for any real product of their own.

(there's a long explanation at http://www.heise.de/newsticker/data/hob-03.08.02-001/
about this)

> I don't know whether it is true, but someone said that
> Microsoft was forced to pay to Symicron because Windows
> GUI shell was named "Explorer", and later for "Internet
> Explorer" perhaps.

Microsoft made a one-time payment of DM 90000 to Symicron as part of an
out-of-court settlement. Basically, Symicron said "Pay, and we promise
not to cause any trouble in the future." Microsoft paid to avoid lengthy
legal proceedings.

Bye,
Christian

Series of interesting email exchanges, nro 0003:

Back in the fall of 1999, I got an email from a man named Marciano Siniscalchi. He had problems using USB modem with B2 and we discussed it ... he was an assistant professor at Princeton university. Nowadays he is a professor of economics at Northwestern University.

A couple of months later, we talked when I was in the process of rewriting the FPU code, it was very buggy and nearly unusable. I already got many bugs fixed but there was a nasty problem with the FREM instruction that I could not solve. But Marciano could! Emails below.

Btw, Marciano quite carelessly gave me a permission to quote him --- I just contacted him and asked! He probably did not realize that this makes him look very smart! Too late to say "no" anymore! Haha!

I'm very careful when quoting other people's emails.

This was typical back then: I got help from many people, from all around the world, and had many interesting discussions!

Lauri

MARCIANO:
    >I *think* I got it. Here's the idea: the quotient in the calculation is
    >an *integer*. Say you want to determine which quadrant a given angle
    >lies in; for example (using degrees), say your angle is 110 degrees. You
    >can use FREM (pretending it works in degrees for argument's sake) with a
    >modulus (divisor) of 90 degrees: you get a remainder of 20, and a
    >quotient of 1. This tells you that your angle lies in the second
    >quadrant, and indeed 7 bits are more than enough to figure this out (two
    >would be enough).
    >
    >On the Pentium, you only get the 3 lsb's, so you can divide the circle
    >into 8 slices (each pi/4 radians, as per the Intel manual); on the
    >68881, you get the 7 lsb's, so you can get more accurate info---you can
    >split the circle into 128 slices!
    >
    >In your example, your quotient of 1.3333etc. should result in the 7
    >quotient bits being set to 0000001, i.e. 1.
    >
    >Hope this works...
    >
    >M

LAURI, WORKING WITH THE CODE AT THE DEAD OF NIGHT:
    You're a genius! I can't believe it, it is just like
    you say! Now Maelstrom works like a charm! "Starry Night"
    (an amateur stellar program) works also. Marathon works
    better, this bug is fixed but some other remains.

    Was that what all of you had in mind all along, but I was
    blinded by thinking about the floating point mantissa?

    Lauri

    P.S. A humble request, may I credit you in the readme?

MARCIANO:
    I would be very happy if you credited me in the readme, but please don't
    call me a genius!

    I also thought that the quotient was going to be a floating-point
    number; I sort of got the idea when I read in the Intel manual that 3
    bits were sufficient to identify which pi/4-sized portion of the circle
    you were in. The comment in the Moto manual is somewhat confusing.

    BTW, you really work long hours! I got your email at 7:28pm. What time
    was that in Finland?

    M
   

Keep it coming, Lauri, please! and before anyone says it, yes, I'm a glutton for Mac lore, lol!

Series of interesting email exchanges, nro 0004:

Alan Cox, the Linux developer and at the time "second in command" after Linus Torvalds, advised this early in 1999. It was the vMac-dev mailing list. I thought that it was a bit too much ... but after one rude email, something snapped and I tried it.

It did not work ... the guy sent me a reply with the suggested subject line!

Lauri

WESTON:
> Lately, I've been receiving e-mail from users who are unacceptably rude.

ALAN:
Feel free to borrow/adapt this

---

Thank you for your annoying email message. This is an automatic post to inform you that while help and assistance is sometimes free, if you wish to be rude and arrogant then the commercial support charges apply.

Your email has been added to the block list. If you wish to get another message past it to discuss this you must start the subject line "I GROVEL BEFORE YOU IN ABJECT APOLOGY:"

---

I find it works quite well

Such people aren't worth dealing with, if someone phoned you like that you'd put them phone down on them, ditto with email.

Alan

Series of interesting email exchanges, nro 0005:

(btw, "nro" was a brain fart; it is a shorthand for "number" in my native language. Since I started it this way, I will continue as well. You can enjoy of this free lesson in foreign languages! Haha)

This story relates to Basilisk II because it was told by a user -- I knew him for many years. I won't do direct quotes this time, but tell the story a bit differently.

I, as a Westerner (well, kind of ... heh), had sometimes trouble understanding the Eastern cultures ... the other day I watched Sofia Coppola's great motion picture "Lost in Translation" with Bill Murray, and it reminded me of this.

And please note, the Japanese guy who told me this, the B2 user, was laughing also ... this is not to put down those great people. I laughed *with* him.

Y, a very nice young man was an enthusiastic B2 user, and I gladly helped to fix some of his problems. Many times I liked to chat about other things as well, or people told me stories spontaneously. Y was presently unemployed, and apparently in Japanese culture, this is a very bad thing: he felt, in his own words, that he was not currently "a member of the society". He had separated from his girlfriend and his beloved dog had just passed away ... playing with B2 offered him great comfort. That was nice!

He told me stories about his engineering career. Previously he was working in a job where his task was to develop PLA chips, PIC microcontroller chips and specialized computer cards. He also designed alarm devices with ultrasonic sensors ... in his own words:

Y:
... most amusing was a molester alarm saying "don't do this" or "you idiot, senile, scum!!!!" since Japanese girls could not say such tough words!

The meaning of this was that the girl or woman had this device in her pocket; when some pervert pinched or grabbed her, she pushed the button and the electronic device spoke instead of her! That was apparently allowed, then.

Amazingly Y continued:

... this product was going to be sold in the USA also ...

What were the company marketing people thinking? I don't know about the USA, but it would not have been a great success in Finland ... If I would pinch a girl on a train or subway over here, she would beat me up in no time ...

Naturally, I immediately replied back to him and asked whether he could sell and send me one of those alarm devices. But sadly, he had none any more after he left the company.

Lauri

Series of interesting email exchanges, nro 0006:

The "miracle request" -- an email subject line.

I thought about this and decided that Cliff would not mind if I told this story even though I have not asked.

In December 1997, Cliff asked my advice about mapping memory under Windows operating systems.

He was very busy; he needed to fly out of town to get one slight medical operation done, he needed to coordinate things to get Executor version 2.0t out of the door, and was worried that the 3 CD's he had promised to me as a "payment" for my help had not yet been sent. Always a gentleman, that was Cliff!

"You're the Win32 wizard. What do you think?"

Haha! Flattering will get you anywhere!

The problem that was bothering him was this: those old and strange Macs placed important global variables at fixed positions on "page 0", the first 4 kilobytes of memory. The DOS version of Executor had access to this memory area, but under Windows, they had to add some offset, then and add and subtract, and that was an extra slowdown. They needed to modify Syn68k, the excellent synthetic CPU made by Mat Hostetter. I'm not sure whether Mat was still with ARDI at that time, he might have already left. Hmm? I never talked to him.

The ARDI guys were very fluent in Linux -- there you would just "mmap" your memory wherever you liked. But that goddamn Windows! Very often there were such random roadblocks, for various reasons.

Together we pondered different options. We considered virtualizing the low memory by writing an exception handler ... but that would have introduced an unacceptable performance hit. I suggested that creating new selectors might help? No...

We agreed how stupid it was that Microsoft was always placing artifical stumbling blocks in the way. It is so brain-dead that instead of including an extra boolean parameter, they chose the option that if some memory pointer was 0, then different things would happen! That is a sloppy, lazy way to design an operating system.

Cliff was worried that I'm wasting my time and suggested that I stop ... he knew that I had a paying job, too. But did I tell you before about my stubbornness ... I was not yet ready to quit, at least until I knew the reason for the problem.

Of course, I knew that there were corresponding Windows API routines that should do the same thing as "mmap" under Unix/Linux. CreateFileMapping and MapViewOfFileEx. But they did not work in this case, for page zero!

This is the Linux code that we tried to repeat. Simple enough:

  addr = mmap ((caddr_t) PAGE_ZERO_START,
      PAGE_ZERO_SIZE,
      PROT_READ | PROT_WRITE,
      MAP_ANONYMOUS | MAP_FIXED | MAP_PRIVATE, -1, 0);

This is the tester coode I wrote for Windows:

GetSystemInfo( &sysinfo );
      mapping = CreateFileMapping(
      (HANDLE)0xFFFFFFFF, // map to page file
      NULL,
      PAGE_READWRITE,
      0,       sysinfo.dwAllocationGranularity,
      NULL
);
if(mapping) {
      mview = MapViewOfFileEx(
      mapping,
      FILE_MAP_WRITE,
      0,
      0,
      sysinfo.dwAllocationGranularity,
      (LPVOID)0 // our zero page address
      );       // test it now...
      ptr = (char *)mview;
      ch1 = *ptr;
      *ptr = 17;
      UnmapViewOfFile( mview );
      CloseHandle(mapping);
}

But no go. So, debugger breakpoints  at appropriate locations and I grabbed my trusty old SoftICE kernel debugger and studied the code to see why this did not work. Just like with cdenable some time earlier.

It turned out that under Windows NT, the reason for the inability to read page 0 was a design flaw I talked about earlier in this story. My email:

EMAIL FROM LAURI:

This code does not work at all under Windows 95, for the reasons we discussed earlier: the low memory area is shared, and when I step into the system code I can see that any reference below 0x40000 is illegal. Trying to force it to work hangs the system.

Under NT, things are different. There is no low memory shared area at all, but it *still* does not work, but for a different reason. Some genius from Redmond has specified that if the last parameter of "MapViewOfFileEx" is NULL, the function behaves differently. That is, the system is asked to give us any linear address. It never occurred to him/her that somebody could want to map to address 0. And, we cannot trick it and specify an address like 1 or 4, since function will fail again - improper alignment.

When I step inside the system code down to the path ,,,
MapViewOfFileEx
      _NtMapViewOfSection
            _MmMapViewOfSection (Ring 0 now)
                  _MiMapViewOfDataSection

... and manually bypass the NULL check, the function returns 0 as it should, and behold - we have the zero page we wanted. We can read from it and write to it.

Conclusion:
- for NT, you *could* have the zero page - with some trickery.
- for 95, you could not.

... and there is no sense in having separate code for 95 and NT.

Well, at least it was interesting.

END OF EMAIL

At first, he was excited that there was some kind progress. But after some more emails and more thinking, it was obvious that Executor needed to live with the extra code to add/subtract offsets. After all, it was no big problem, the program was fast enough anyway. Perhaps someone else could have solved this problem, but I could not Obviously, I was not up to the task to do miracles! Haha

Lauri