DAILY: Saturday, June 16, 2001

read book now





An irregular journal of things computerish.

For the BYTE story, click here.

Previous Weeks of The View 1  2   3  4  5  6  7   8

atom.gif (1053 bytes)

If you want to PAY FOR THIS there are problems, but I keep the latest HERE. I'm trying.

This is a day book. It's not all that well edited, and some of it will be repititious and boring. If you want the original CHAOS MANOR columns, stay tuned: I'll take this mess and make a real 4,000 word column out of it monthly, at least for a while, since the overseas BYTE magazines seem willing to buy translation rights. Stay tuned on that. In fact, see below:

I have am still going mad over making up a mailing list. UPDATE: FLASH: Mailing list problem SOLVED. See Thursday entry in VIEW 8.

I know you mean well, but PLEASE DO NOT send vague references to bad links. If you want to tell me something is wrong, please be specific about what link on what page. I got one the other day that I ended up wasting half an hour looking for, before I came to my senses...

Monday Tuesday  Wednesday Thursday  Friday Saturday  Sunday

read book now





Monday, July 27, 1998

This has to be fast. I am off to meet the producers of Buffy The Vampire Slayer in an hour or so. That may be interesting.

Regarding Winnie: that MSI mother board like many doesn't really have on and off: instead, there's a software system so if you don't hold the off switch down long enough it "sleeps." When that happens there's power to the mother board, and it won't forget the date. Even if it is OFF there may be some kind of trickle power. So last night I removed the battery and unplugged the machine. I am pleased to report that this morning it didn't remember a thing. Had to reset the date.

Regarding fdisk: because the machine was so flakey, I scrubbed off the Windows directory entirely. Then I tried to reformat. The result was horrible. Apparently W 95 b even when you don't choose their fat-32, does interesting things to a disk format. The result is that DOS 6.3 fdisk doesn't work. I'll see what I can do about that when I get back from my appointment. The idea I had was to start all over on this machine, and I will. I'm using W 95b because I want to install the Ricoh MediaMaster CD R/W and that won't work under W 98, at least not without a great deal of reblowing of proms and stuff.

Regarding Partition Magic, if there's a version that installs from DOS I don't have it. It wants Windows to install it. Since what I wanted to do was make sure the disk partitions were all right before I installed Windows -- given the unreliability of the system prior to that, with messages saying I had to re-install Windows and such I wanted to be sure all was well -- I sure don't want to install Windows before I deal with disk partitions.

What I have is Winnie with DOS. I have to boot from a magic floppy to make it recognize the CD ROM. I can put the drivers over onto the C drive, and I may. My goal is to be sure the disk partitions are all right, then reinstall Windows 95b, and this time add the Creative PCI64 board and the SOHO network board one at a time, be sure all that is working, and then see if the SOHO will do 100 megabits. Since the Diamond Monster Sound board had left of traces of itself in there, it may have helped corrupt the registry. In any event I can say with some authority: DO NOT USE A MONSTER PCI SOUND CARD IN A WINCHIP MACHINE (It's good with many other chips), and if you do, don't merely remove it and use device manager to remove the drivers. UNINSTALL it.

I will see if I don't have a later version of fdisk; if not, the system looks clean and after I get back I'll put up Win 95 b, install Partition Magic, and see what things look like from there.

I also have to write an essay for Intellectual Capital, and get my monthly column made up out of all this mess. That sounds fun…


It was a good meeting. Niven and I may do a TV series. Or a movie. Or both. We have some great ideas. More on that later.

Now to deal with the WinChip system. First, Peter Glaskowsky, my favorite chip expert, tells me that overclocking to 75/225 is doomed to failure. His full letter is in mail. Anyway, I should cut back to 66/200. I always thought that myself, but it was worth a try. Apparently it wasn't worth a try.

Now to see what I can do reliably with a WinChip. One thing is to crank that clock speed back to 66/200, and I think it tells in the mother board book how I can do that…

Yes. It's one dip switch. Leave the 3:1 ratio, change basic clock from 75 to 66. Boot up. Boots fine.

Partition Magic can run directly from the CD, no graphic displays. Says both disk partitions are fine now. OK, installing W 95 b again…

And Later still:

The first thing I notice is that everything is actually faster. Apparently there are fewer retry errors in copying, which is all I have done, but copying all the W95b files from the CD to C:\Windows\Options\CABS took FAR less time with the slower clock speed than with the faster.

We will see, but I have a much better feeling about all this. For the record, this is a 75/225 WinChip running at 66/200. I have other mother boards and I'll try again with another later.

Now to install W 95b. I have a good feeling but I think I will still do it without sound board, then add, sound board, then add network board, then add SCSI and Ricoh MediaMaster. Proceed with hope and caution…

My only real question is do I install the Monster board at the lower clock speed and see what happens, or do I do the PCI64 first? I think the latter. Stick to what I am pretty sure will work, run at that for a bit, and take excursions off a known base. That's how we used to do the X programs in the aerospace business, and it's a strategy I know...

Progress report:

I never had an easier installation: everything is going smooth as silk. The only minor problem is that the system got just a little confused by the Teac 6 changer, but that was in part my fault, and none of the changer software has been installed: Win 95 just found the 6 drives. The confusion came as to which drive had the Number 9 software when I did "have disk"; took about 20 seconds to straighten that out. Otherwise the EIZO monitor and the Number Nine card software went in as smooth as ever I saw.

Now to continue with other stuff, but so far, WinChip at 66/200 is working just fine.

Continues smooth. Shut down and restart work fine. Utility software installs fine. There's a solid feel to this machine. Now for a sound board. I'm tempted to the Monster, but let's try the Creative PCI64.

Incredibly smooth. The longest part of the installation was IE4; Number 9 had installed IE 3, and it seemed reasonable to let Sound Blaster upgrade to 4. Otherwise it was very fast, and on restart I get a blast of sound. Of course we'll have to see with games programs and if I get lockups, but it looks good up to now. Real good.

Oh, crud. In letting Creative install IE4 I also let it install Active Desktop. Not sure what that will do for me, but I am pretty sure I don't want it. We'll have to keep looking for just a while, but I don't need my resources used up with that kind of pretties.

It took about two minutes to decide I don't want Active Desktop. There seems to be no way to remove it other than uninstalling Internet Explorer, so I have done that. We will see what happens now. I may have to reinstall Windows 95 because I made the mistake of allowing Creative Sound Blaster PCI64's software to install Internet Explorer 4 and somehow I got "Active Desktop" with it. I wasn't paying enough attention, I guess. Active Desktop seems thoroughly useless. If there are those who like and use it, please explain what it does and why I would want it. Incidentally, it REALY doesn't want to uninstall IE4. It's protesting all the way. It did do the job, though, and I'm restarting now. We'll see if Sound Blaster works all right.

It does. So does the SOHO Network card, at both 10 and 100 megabits. Everything installed properly and easily. It's pretty clear: running a WinChip at 75/225 is not a good policy. The same chip at 66/200 seems solid and so far works just fine. I now have a fast Win 95b system. I'll see about adding SCSI assets to it shortly.

I should install IE at some point, but without Active Desktop. Active Desktop seemed both unfamiliar and, at least as default, seemed rather childish, with needless graphics, the words "My Computer" in the middle of the "My Computer" display, some icons large, some hidden, things hard to find. I may be reacting badly to the surprise of seeing it unexpectedly.

In any event the first WinChip adventure is over. In a normal column you wouldn't have seen all the false starts. Oh well. The conclusion is, Don't Run Winchip at 75/225.

Midnight: extensive tests with non-well-behaved programs and games. No glitches. WinChip with Creative PCI64 Sound is stable and reliable when WinChip runs at 66/200; none of the problems I had at 75/225. So it goes.

Tuesday, July 28, 1998


It's late, it's hot, and the air conditioning system has completely ceased to function. I have the case off Winnie and a big fan blowing directly on the chip fan; started that when I was getting flaky responses during the overclocking episode. Winnie is working fine, but I am out of reliable SCSI boards, so I can't do any more tests until I get a couple. Winnie ran all night, and did a number of complex scenarios of "THIS MEANS WAR"; those were almost guaranteed to cause glitches in any system because of the number of parallel events, but they ran smoothly, and the sound was smooth. Winnie may be the most reliable games machine I ever built. I'll keep working, but running that WinChip at 66/200 is the right answer.

But it is late, and I am lazy, so I'll let someone else work for a change:

Eric, our long suffering Chaos Manor Associate, has been using Windows 98 for a long time, and gets a bit exasperated with me. After the active desktop mess, he said:


The appearance and behavior of the system running IE4 is easy to alter.

In the Display control panel there will be a new tab marked Web. This lets you alter or disable the active desktop and offers a link to another panel for controlling folder bahavior. In Win98 this stuff can be reached directily from the Settings section of the Start menu.

It's all there in plain sight if you look.

Which is all true, but what I replied was, "Well, I may try it again, but why do I want it to begin with." Of course I got a real answer to my flippant question.

It’s one of those cases where Redmond demands the code reside on your drive. They’ve made up their minds and there will be no further discussion. At this point I guess they start waving Nun-class steel rulers to drive the point home. At least until IE5 which is supposed to pull back from some of the more annoying elements of Active Desktop like the Channel Bar.

The stuff will be sitting on your drive if you install Win98, which is inevitable for the purpose of what Chaos Manor does. The reasons for its existence raise some questions in marketing philosophy. Should new users repeat the learning curve earlier adopters went through or should things get easier. The number one reason for first time computer purchases today is Internet access. The Web has become so central to that experience that many do not realize it is only one layer of the net. So why not suck up to new users by making everything look like the Web? The existing users won’t mind. They demonstrated in the last twenty years they’ll tolerate just about anything.

That Ullman article irritated the hell out of me too when it first appeared. Apparently, being a ‘professional programmer’ in her world is possible without even a rudimentary concept of what’s in the box. For instance, the place where she becomes fascinated by the ancient hooks to the ROM (or Cassette) BASIC interpreter in the original IBM PC. This is like an architecture student obsessing over an olddisconnected sink pipe in the basement while ignoring the cathedral above. The project she refers to failed for reasons other than programming techniques. Her description makes it obvious that the project leader had no idea what skills were needed where. The finest ActiveX (nee VBX nee OCX) components in existence won’t help if the crew doesn’t understand the code that is unique to the project. The generic bits didn’t make or break anything.

The installation saga she describes sounds like the fumblings of a first timer operating under any number of misconceptions. Get rid of an OS, any OS? Move data to be saved to a safe location and format the partition. Bang, it’s dead. I’ve never been a professional programmer but my limited knowledge has let me install three OSes on the same machine in a single working day. Given, I was familiar with install procedure for each OS but making the previous install go away was never the stumbling point.

At the time Ullman’s article first appeared I found myself in an argument over the ‘evils’ of prepackaged components. My opponent lived in an elitist fantasy land where he could claim complete understanding of every single line of he wrote. When would such a person ever find time to write any new code after spending every waking hour divining every facet of a modern OS’s source code. The funniest part of this was that my opponent is a big Mac fan who neglected to notice how dependant he was on the Mac ROMS. A huge part of the whole Mac concept was this body of code for programmers to consistently use instead of reinventing the GUI with every project.

Another good example is device drivers. These can be the bane of our existence but would anyone want to go back to days when application designers had to roll their own instead of letting the OS and hardware companies be responsible. A bad driver can be a pain but at least it is consistent. Remember when a high-end video card’s ability was only reflected by the handfull of apps that had drivers?

This is not intended to downplay the advantages of open source but the fact is a productive programmer has to trust in a large amount of code written by others. If your task is to write an expense tracker for field agents how much time do you want to spend writing support for a zillion printers, video cards, sound boards, etc? Chances are the guy doing business apps for CPAs isn’t qualified to write hardware drivers in the first place. It isn’t his job nor should it be. Regardless of source code availability he ultimately has to trust others to do their jobs.

And continued with:

A few thoughts to append to the previous message.

I've read your Escape From The Planet of the Apes and I think it would be obvious to anyone you tried to make the most of the material. I recall you addressed several issues the films glossed over.

That reminds me. Have you noticed that the press has obsessed over the discovery that the guy who shot up the Capitol Building had, gasp, science fiction literature in his truck. What would they have said if he was a Clancy reader? The Jack Ryan series has had more elected blood shed than any SF novel I can recall.

It is alarming that the Navy has placed such critical functions in the hands of NT. I find it even more frightening that the vessel cannot quickly recover from crashes. I wouldn't have used NT for this purpose but I find some of the implementation questionable just the same. Why would a mission critical system ALLOW invalid data to be entered? Especially something as excruciatingly common as creating a divide by zero situation?

If I were designing that system I would want to use something like QNX's offerings to create a system that can be put mostly in ROM on a single board or a small box. It would be pretty stupid to lose a battle because a single shot killed a PC. You can't call a time out while someone runs over to Fry's for a replacement. I'd want an armored cabinet full of battery powered duplicate systems that can be slapped in place quickly. Panasonic has a line of hardened 'industrial' laptops that could fill this role. An fallback wireless network might be good idea, too. Again, imagine losing a battle because an Ethernet cable got severed. Rather more difficult to overcome than replacing a cut intercom with a chain of hollering crewmen.

Which says it as well as I could. Given that my son Phillip is a ship driver and will probably command an Aegis cruiser one of these days, I have some interest in these matters…

Another comment on Active Desktop from Robert Maxwell []:


I simply deactivate Active desktop most of the time by right-clicking the desktop, choose active desktop, then make sure view as web page is deselected. My wife is a meteorologist, and I'm almost one, so when storms go through our area, I turn it on, because I have added several weather radar web pages to the desktop, and it will update them (although the update scheduler is rather inflexible and could use work) regularly, so I can keep an eye on things and know if there's going to be something fun to watch. Of course, I do this from work, with a fast internet connection. I wouldn't even think about it over a conventional modem. The channel features blow donuts, as do most of the rest of the "web interface" of IE 4 and win 98. IE 4 all by itself is not a bad browser, although the cutesy names get annoying. It does start and run java better than netscape.


Finally, Eric on Office 2000:

One of the news services surprised me by knowing it was my birthday and supplemented that surprise by mentioning it is also Elizabeth Dole, Thelma Todd, Peter Jennings, and Stephen Dorff’s natal, too. Also the signing date of the first US-Japan commercial treaty, the first motorcycle race, the first nonstop transpacific jet flight, an earthquake tha killed almost .25 billion Chinese, and (ugh) the wedding of Charles and Diana. Some significance I could do without.

I don’t know if you’ve already met with the MS Office folks but some less immediate ideas have bouncing around my head about how to make PCs more dependable. It’s become apparent that there is little Lotus or Corel can do to change the course of the PC application market. This can be seen as good for users who only have to learn one set of apps but bad for driving improvements in the code base. The normal reward and punishment cycle that drove development when virtually all users were technically oriented is now broken. The average can’t properly voice their complaints because they don’t know what to ask for. As a result every new edition has lots of new code for new features but tons of bad old code because nobody expects a reward for fixing it. It’s difficult to judge the performance of the newest OS releases because it’s almost impossible to create a real world situation that doesn’t involve clunky old code that was written before the OS’s important features existed. The only hope I can see here is for Microsoft to go into competition with itself.

Many would like to see Microsoft rewrite Office from scratch. This simply isn’t going to happen. The project is too massive to have any hope of fitting into the normal update schedule. But most of us don’t need Office in its entirety. 90% of our time is spent using using only 10% of the functions. Millions of users could be very happy with something that offered the stuff they needed and read the existing files while offering a small footprint, low resource usage, and most importantly, stability. So, lets have Office Lite. A new suite written entirely new incorporating the best Microsoft’s huge talent pool can create. Apply everything learned along the way while disposing of the dross. I’m not suggesting that everybody should be in on the project full time but only that anybody who wishes to offer analysis in their spare time be encouraged to do so. Call it corporate-level open source. The project operates openly but not chaotically. One of Linux’s big advantages in the eyes of many is its lack of a single standard interface but this is also a major reason it fails to be viable for consumers who don’t want to deals with myriad desktops each with it’s own idiosyncrasies with applications. Office Lite must have strong management who knows when to lock in certain items.

So we have two Offices. One for power users who don’t mind enduring legacy issues to get more features and other version for normal folks. Office Lite would probably benefit if only runs on NT and WinCE systems, ignoring the 16-bit warts in Win9x. After two or three generations Office Lite catch up with or exceed regular Office in features. At this point Office Lite becomes Office and a new Office Lite development cycle starts from scratch.

Maybe I’m nuts but I can dream.

I have little doubt that WinCE will start finding it’s way into a new class of desktop systems. Remember when you could completely blow away an Atari ST’s hard drive but the system would still boot because the bulk of the OS was in ROM? ROM and upgradeable FLASH ROM has made huge strides in low cost and capacity. For example, the highly anticipated new title in the Zelda series coming this Christmas will be a massive 32MB for only $60. Hitachi is preparing to introduce Type III Flash cards with capacities of 300MB and greater with the price per megabyte rapidly moving toward the $1 point. These prices may seem high compared to CDs and magnetic drives but these only be needed for the core elements of a highly reliable system. The cheaper traditional media can be used for data storage and less than critical applications. There is no reason we couldn’t have a low priced modern system with the dependability of the ROM oriented machines of yore.

The first step is to draw up some priorities. What niceties can I live without versus what components will put me out of business if they fail. Text editing is critical but I can live without a font library for a while if something happens to my hard drive. If I had to chose only e-mail or web access for a week the e-mail would win with very little consideration needed. Most of these decisions aren’t difficult.

In cooperation with this concept apps should be designed at two levels:

A solid core to reside in a ROM partition and less critical features loaded from disk as needed. If my drive gets blown away I’ll do without the grammar checker happily if my indispensable editing functions are intact. Applications should also be well contained. The way Windows currently scatters bits and piece all over the drive is disaster waiting to happen. Disasters will happen, though, so why not take advantage of extremely low cost of drives to separate application code from data. Damage to one should not threaten the other.

I think most consumers would be happy to get off the PC merry-go-round in favor of something that offers the same power without the hassle of the hobbyist box that has become the standard for all.


Afternoon. It's hot. I worked on Mamelukes in the Monk's Cell, which has its own air conditioner, so I didn't realize just how hot until I got up here. It's really hot (over 100 F), and my air conditioner is dead as a doornail, and I can't find anyone to fix it.

I wish the Framers of the Constitution had known about air conditioning. They could have made it unconstitutional to have an air conditioner of any kind within the boundaries of the District of Columbia. That would get the government out of town for the summer, which would give us two or three months less government every year. Oh well.

Winnie now has the Ricoh Media Master CD R/W drive in operation, using Adaptec DIRECT CD. A small tale goes with it: the only SCSI board I could find was an ISA board. I won't leave it in the system since it slows the system down a lot, but I did want to test the CD R/W.

When I first put the board in I didn't pay a lot of attention to IRQ so of course it didn't work: IRQ 11 is in use by the PCI Sound64, which seems to have grabbed several interrupts; I need to look into that. Why does it need so many? Anyway, Device Manager soon told me I'd have to change to IRQ 10, which I did: and on reboot the system locked up. It took me a bit to figure out why: in the AMI BIOS for Winnie, IRQ 10 was set to PCI. Once I went into the BIOS and set IRQ 10 to be ISA, everything worked fine. One tends to forget these things…


The language debates continue:


Steve R. Hastings []

I am a professional Software Development Engineer. I have been working with C and other languages for about 15 years. I believe I am qualified to evaluate C as a language.

C is heavily used because it is the best language of its type. Pascal and

Modula-2 lost the competition because they are inferior. ANSI C has stolen

all the good ideas of Pascal, with none of the headaches. What sort of

headaches? Read this:


C has its warts, but overall it is a clean language that makes it easy to write good programs. It compiles to very efficient code, and is portable across an amazing array of systems. Pascal and Modula-2 don’t compare, let alone BASIC. Never forget that when you are using Visual Basic, you aren’t using BASIC; Visual Basic is unique and utterly nonportable. I simply can’t use it for the projects I work on. Pascal is also very non-portable;

I quote to you Cooper’s Law of Standards, "If it doesn’t work, it doesn’t stay standard very long." Turbo Pascal, from about 3.0 on, was also a very nice language—and it wasn’t portable because it wasn’t Pascal. C, and now C++, really is portable. And that’s important.

Sure, there are C programs that are horribly written. That’s not C’s fault. There have been languages that attempt to force people to write clean, maintainable code by forcing them to work in a straightjacket. These attempts always fail. Good design can’t be coerced, and good design is more important in reading and understanding code than anything else.

It is no more fair to say that C causes bad programs than it is to say that English causes bad novels. There are many more bad novels out there than good ones, and so it is with computer programs. I’ll cheerfully submit any of my C programs for your perusal; you can decide for yourself whether they are innately hard to read.


Steve R. Hastings "Vita est"

A good statement of position,. I suspect I have not made myself clear.

One does not judge a nation's literacy by the achievements of its best writers. A nation of illiterates can produce great literature; it doesn't make the nation literate. For the same reason, one does not judge the readability of code and programs from examining the very best or the very worst: one must use the average, the day to day yard goods; and on those standards, C fails miserably. There may be readable and maintainable large programs written in C, but they are rare. The average big C program can't be maintained by anyone who wasn't in on writing it, and often not even by the original authors.

Hastings's title reminds us of Dijkstra's essay "GOTO SEEN HARMFUL." The original is worth reading, and applies to C as much as to the languages Dijkstra was writing about at the time. C allows too many liberties. A C compiler will compile utter nonsense: the result is that it's easy to get a program to compile, but it doesn't do what you wanted it to do without a lot of debugging. That's the wrong direction. The goal ought to be for the compiler to catch all the errors including errors in logic so that a compiling a program might be difficult, but once it did compile, it should do pretty well what you think it will do from reading the source. Pascal, and Modula, were headed in that direction.

Now it's true that Modula-2 didn’t get the refinements and attention it needed to become an important commercial language, and thus I have pretty well lost the battle: but my point was that if we are going to have C as our major computing language, WE WILL HAVE NO CHOICE but to go down the road of software wizards that so upset Ullman. If you choose tools that require a Hemingway or even a lawyer to write a simple business letter, you will never have anything like universal literacy in your population. C requires programmers with the skills of a Steve Hastings, and there are not enough of him to satisfy the need for programmers.

ADA was a step in the right direction, and should have been continued; the problem with Ada was that too many people got in on the act, and it was loaded up with exception handlers, bells, whistles, mangers to feed donkeys, and the kitchen sink. It could still be recovered. Or Modula could still be turned into a real language: it needs libraries and code generators, but that's not impossible even now.

As to C's famous portability, that's not as important as it once was. Computing machinery is getting cheap, and programs that can take a source and translate it into anything you like are not that hard to write. C++ is a case in point, and attempts to make C into a more readable and useful language. I don't think it succeeded.

I don't really have a remedy here, and perhaps there is none. I attempted two points: a lament that we went down a road toward a language that demands, no requires, great skill in order write readable programs while allowing, no encouraging, average programmers to make use of tricks whose purpose is soon forgotten; that will happily compile nonsense. I wish we had gone in the other direction, toward tools that demand clarity of thought, and produce source code easily understood. We didn't go that way, but I can still lament it, and hope we may some day go down a different road.

My second point, I repeat, was that given that we demand code be in C, we will find ourselves using more and more software wizards: among those I include Visual Basic, which is certainly as much wizard as language, but which make it possible for school teachers to do things that top programmers couldn't do a few years ago. Wizards and wizard-using languages like Visual Basic have their own problems, and certainly are not efficient of computer resources: but then computer resources are cheap. When computers were more rare than talent, it made sense to conserve computer resources. Now, computers more powerful than early supercomputers are available for a few hundred dollars, surely it makes sense to use computer resources to conserve talent for situations that really need it? Making Hemingway write thank you notes because our language is too difficult to allow anyone with less talent to do it makes no sense. I trust the analogy is clear?



Wednesday, July 29, 1999

I have the Ricoh MediaMaster CD R/W working on Winnie, but there's a hitch: Adaptec EZ CD CREATOR will not recognize the IDE TEAC 6 changer as a CDROM source, so I still can't copy CD's, at least not directly. The main reason Winnie is Win 95 OSR2 and not Windows 98 is the Media Master won't work with 98. I CAN copy CD's: I copy the contents to a directory on the D drive -- it's nearly empty anyway -- and then make a CD with that content. That works, but it's mildly annoying that I can't copy direct. I installed EZ CD Creator both from the Ricoh and from the TEAC IDE; installed from the RICOH, EZCD Creator can't even see any other CDROM drives. Installed from the TEAC it sees the 6 drives, it just doesn't believe they are sources. It may be that you must have both source and destination as SCSI drives; I'll ask since the Adaptec people are coming here this week.

My next task is to build a WinChip machine for Windows 98. I need some more memory to do that. It's cooler today.

Eric makes this observation concerning the code wars.


I’m all for readable code but I don’t think the need for Wizards is a product of languages like C or its offspring but rather the environments of modern applications. Most Wizards are to quickly implement functions that can be genericized and gotten out of the way quickly to concentrate on what makes the project at hand valuable. If you’re a CPA turned programmer whose task is to simplify the live of other CPAs, your value comes from understanding of their work and less from a mastery of graphic techniques. If a spreadsheet layout is the answer for a given problem why write your own grid functions when many excellent COM controls that do the same thing are available for pocket change?

Perhaps the point of contention would be reduced if those prepackaged components included source code for the benefit of those inclined to delve. Since the people who benefit most from components are specialists first and coders second this may not change much but we can hope for faster bug fixes and better customization. The question is whether component suppliers can feel safe essentially exposing their trade secrets to the unscrupulous. If the value of your product derives from mastery of quality coding for a platform it’s understandable that you don’t want to leave the ‘warehouse’ unlocked for any thief to just grab some merchandise and sell it under their own label.

This is a sticking point of the open source approach. It assumes opportunity after the sale. A Linux specialist would derive his profits from service rather than the disc but what about products that don’t require service like entertainment and education titles. A game that sold for $5 but accrued many times that amount in support costs would greatly defy any definition of fun in my dictionary. Likewise for something like Mrs. Pournelle’s TLC. This would end up teaching the instructor instead of the students and teaching them the wrong subject to boot.

All to many overenthusiastic Open Sourcers trumpet the idea that all software should be free. I think this is very shortsighted and could prevent Linux from achieving a consumer market by scaring off the companies that develop for that segment, not to mention the retailers who stock the products. (We can dream of a day when all bits are distributed over the net but don’t hold your breath.) There needs to be an understanding that not all products can be sold the same way.

All of which is clearly true. Well said. I've been trying to hammer that theme home for 20 years: the computer revolution won't be finished until it's more important to know what you want the computer to do than to know how to teach the computer to do it. I have always believed that if we'd continued down the Modula/Ada road, we'd have moved toward computer languages that were themselves loaded with wizards (read: libraries of functions) making it possible for an accountant to learn a powerful programming language in a matter of weeks. Visual Basic attempts that, witness Mrs. Pournelle's efforts to learn it so that she can put her reading program into Windows. (And yes, I could do that, but we have to make a living too; better she learns to tweak it herself, and I think with VB she can; I wouldn't expect her to learn C.)

Does anyone recall RATFOR? This was a precompiler for FORTRAN that allowed declarations and rational structures (RATFOR = Rational Fortran). This was an interesting development which I liked, but which seems to have vanished down the C rat hole. Lest anyone be unclear, I don't question the motives of those who believe the way to go is arcane languages understandable only by experts and then only by experts working very nearly full time at it: but I do note that their opinion and their interests are congruent. It may be that's the best way to go, but I remain unconvinced, and Ullman's laments about the horrors of wizards don't help.


Just looking over the ZDS utility site. Several of those touted are programs I didn't bother to review as being not worth the effort. The late S. J. Simon's masterpiece book on contract bridge (Design for Bidding) said of bridge bidding conventions that how well they work for their own purpose is the last thing to consider: first is what effect do they have on other hands where they are not going to be used, but have taken up a bid that one would want to make in an unconventional sense. Yet, he said, there is a sense in which that's the first consideration because if they don't do something you want done, there's no point in looking at any other aspect. There are a couple of utilities in that list I put in that category: even if they work, why would I want to do that?

Drop Chute is an example: as we saw it demonstrated, it locks your system to someone else's under certain circumstances. The only way you can get control back is to call the other guy and get him to log on so that the transaction can be completed. Or reboot. I am not sure I need that. They may have fixed that bug: Darnell and I found it while playing with their demonstration on a show room floor; but even if it is fixed and it all works flawlessly, I couldn't see the point.

There are a couple of others like that in the list. Oh. Well.



I think I quit. I think it is time to write novels, or even work with committees doing TV and movie scripts. Nothing can be as frustrating and time wasting as these computers.

Eric was over today and we installed my Microsoft Force-feed joystick on Winnie. I found that I am as unable to play those games as I thought I was, so I shut down. When the machine rebooted it locked because there was no disk in the CDROM. I shut it down.

Now Device Manager has yellow exclamation points in the IDE drivers, and there is no CDROM. None. If I boot in DOS I can see the CD ROM. You cannot remove the IDE device drivers. I can't find any new one. Nothing works. I may have to reinstall Windows. I am not sure how to do that even. Delete the whole damn Windows directory? What? I am furious. There appears to be no remedy, and I don't even know what we got wrong. Some kind of drivers but what? The Network works fine. The C and D drives work fine, But it will not believe there are any CDROMs or ZIP drive on the system, and Device Manager is certain that the IDE drives are not operating properly but I cannot remove them.

So now Winnie has no CD ROM (other than the SCSI RICOH and I removed that one actually since I thought perhaps that was blocking it.) I have told the BIOS that there's a CDROM on the Secondary Master. Nope. And the ZIP Drive has vanished too. I hate this. I hate this. I am weary unto death of this. Better to write books. Better to sit in meetings with producers.

Does anyone remember which are the stupid files you must remove in order to reinstall Windows 95?

Doesn't matter. I have deleted the entire directory. Why not?

How much of my life have I spent waiting for a computer to reboot? Watching a blank screen? Looking for a pencil and paper so I can write down the serial number of a CDROM? Enough, I say. Bloody ENOUGH. Better to take meetings and do lunch in Hollywood, something I have avoided for 30 years and more. But something snapped a minute ago. Why should I have to put up with all this? I don't get paid much to do it any more. It distracts from the rest of my life.

I HATE this, and I am beginning to hate these machines. How in the world can you get the IDE devices so corrupted that Windows can't find them? Why do I have to reinstall everything? And this over a stupid JOYSTICK? It can't possibly be worth it.


Hours later, it works again. This is silly…

Thursday, July 23, 1998

Last night Claude Addicott told me my problem might be bus mastering. This exchange followed:

Dear Dr. Pournelle,

"Tell me more about bus mastering. I reinstalled. That worked but what a pain. Hmm. If adding mscdex would do it—the problem was that device manager thought the ide devices were broken. Yellow exclamation points. And you cannot remove these, so you can’t let plug and play fix it."

Usually when this happens you first get a blue screen error message from Windows telling that there is a problem with some devices and to Reboot. There are actually several ways to fix this. One involves using Regedit, but the easiest is to install the Bus Mastering patch for your motherboard’s chipset. (Intel’s patch is available from their page, I have it and several others). This patch is also supposed to speed up the hard drive access. I never saw it with Intel, but on my VIA chipset it did make a difference. What the yellow exclamation points on the IDE devices mean is that the 32 bit drivers aren’t working so Windows is accessing the devices through DOS, that’s why you need to use MSCDEX (and the DOS driver in your CONFIG.SYS) to access the CD-ROMs until the problem is corrected.

"Reinstall works. What fun"

I know, that’s what I did until I found out about the Bus Mastering patch.

BTW you didn’t need to wipe out the Windows directory to reinstall Windows (even with a version that is NOT an upgrade). All you need to do is delete (or rename) WIN.COM. Also when SETUP asks you where you want to install Windows, choose Other Directory and type C:\Windows (otherwise it tries to put it in Windows.000). For a new machine like Winnie it probably doesn’t matter, but on one you’ve been using a while this lets you keep all your settings and shortcuts.

If you let me know what chipset you motherboard uses (Intel TX, TXPro,VIA VP3, ...) I’ll try to send you the correct Bus Mastering patch.

Claud Addicott

I do know that you don't have to wipe out everything to reinstall Windows, but it's about as easy to do that as anything else. I have the cabs files on the D; drive so as long as I don't have to reformat the disk it doesn't take all that long to install. Just annoying, and I got to wondering just how much of my life I have spent staring at blank screens while systems reboot. Too much of it, I think.

Overclocking Winnie was an adventure. I don't mind adventures. What I don't expect is to have an adventure when installing a Microsoft powered joystick and the software that came with it. THAT was an unexpected expedition I could have done without.

While I was at it, I also wanted to install and play Privateer, an old Wing Commander offshoot. Alas, it's nearly impossible. I'd have to make up a DOS boot disk with QEMM or some other memory manager, and the CD ROM drivers including mscdex, and the various sound drivers. I vaguely remember how to do all that for making the system boot up in DOS, but again, just barely. A lot of trouble to put up a years old game I have played through to the end several times, and I probably won't bother. Pity. Privateer and it's add-on scenario Righteous Fire was one of the best games ever done, with a story line but a lot of free-form adventure capability. I wish they'd release the scripting language to let us put together our own stories in that universe. Instead, Origin came up with the impossible Privateer sequel, which was real cool graphically but sucked dead bunnies as a game. One of these days I'll do an essay on games designed to impress other game designers rather than the games players…




Hiya Jerry...

I thought that I would drop you a line and tell you what the problem was with Winnie when you had yellow excamation points and no access to the CDROM.

(I found this out after wrestling with this problem for weeks).

Sometimes Windows95 tries to use protected mode drivers on one IDE channel, and real mode drivers on the second. On certain chipsets, Windows 95 will, at that point, freak out on reboot and lock. At that point, it will add NOIDE into the registry under the registry key:


and then refuse to use protected mode drivers again. This has the effect of disabling all ATAPI devices like CDROMs and ZIP drives, as well as 32 bit hard drive access. If you delete DOS drivers from autoexec.bat and config.sys, and then remove the NOIDE using regedit from that key, then the IDE drivers will reinitialize, and you should have access to those drives again. The Microsoft page that describes this is:

Just in case you ever run across this problem again (I have seen it happen to me on three computers so far).

Love the page...keep up the good work.

Joe Damiani

Network Specialist

Scientific Atlanta Canada Inc.

Thank you. NOW I understand what happened. And Regedit ought to have done the trick. Now that I know what to look for. Thank you.



Saturday, August 1, 1998

New Adaptec EZ CD Creator installed on Winnie. Can now copy CD's with the TEAC 6 changers as a source. There's a lot more software for cleaning up old scratchy records -- you can make your Caruso 78's sound pretty good now -- and a great deal of software for creating slide and sound shows, where you have a narrative in your voice and at the end of it the picture changes. There's also a lot more software for dealing with pictures, and making a self-contained disk that will display pictures when played without regard to whether the images are jpeg or giff or whatever. Needs testing, but the demonstration was positive.

Still some minor problems with Adaptec's "DIRECT CD" which is the software that lets the Ricoh R/W do R/W as well as record. According to the latest projections everyone will soon buy CD R/W rather than CD/R even though there's no great reason to. It costs very little more to make an R/W drive and for the moment you can sell them for more at a profit. Why people want CD R/W rather than just CD/R isn't clear to me. The R/W blanks cost more (although that price is falling fast). On the other hand, they may eat M/O disks alive. I like M/O with 640 meg disks, but R/W blanks cost no more now, and everyone will soon have an R/W drive, while M/O never really caught on big. Oh. Well.

Saw Mask of Zorro last night. Predictable movie, but generally a hoot. Very underplayed sex scenes, which is to say, it stayed romantic and the only sexual relations are monogamous and very much off camera. The fencing scenes were actually pretty good, not as wild as Fairbanks, and more realistic than I'm used to. The villains are properly villainous, and while the plot is a little incomprehensible -- how Don Rafael, once a Spanish governor, returns as a governor under Santa Anna during the US Mexican War isn't understandable to me -- it doesn't matter. The Bad Guy is Back, and so is the Good Guy…

It's a nice day. My air conditioner is fixed -- turned out to be a bad controller. Got a new controller which I must say I don't like as much as I did the old one. It's digital, and the switches aren't as accessible, and I think the thermometer is WAY off. I need to get a more accurate thermometer to calibrate it. But at least it works, only now the heat wave is broken. But that means I can go hike in the hills, and I'm off….

Well, slight delay. One of the most irritating things about Front Page is that if your connection to the web site is broken at any time -- i.e. if you hang up your modem connection -- then Front Page cannot open any page for editing. All those pages exist on my local disk, but it can't open them if, at any time, it has lost connection to the external site. This is STUPID. It may be some imbecilic setting, but if so I can't figure it out. All I know is that if at any time, no matter that Front Page has been in background all the time, I drop the connection to the web, then reconnect, then the next time I ask Front Page to open a file it won't be able to do it. I have to shut the sodding thing down and open it again. This seems like one of those design features that make you want to find and beat senseless the designer.


This came in last week. On investigation the bug has never been observed at work -- like most of these 'security problems  it was found by a team of bug finders -- but it is probably worth doing something about.


I was listening to NPR while driving home from Computer &; Software Outlet with a load of Pentium II's, memory, etc., when a report about a serious security flaw in OL98 came on. You can read about it off the main Outlook page at There's a small patch file available for download.

I figured that the whole world knows you use Outlook and you're a prime candidate for getting hit with this bug.


Robert Bruce Thompson

I've done the download and patch. I suggest you all do the same.


And a note from Frank Gasperik, who has appeared in various stories by Niven and Pournelle in various guises, most recently as Hairy Redd in Footfall. He suggests you look at

where there is clearly a serious problem that the World Health Organization can't solve. And what about the children?

This Front Page problem is serious. I made the hideous mistake of working directly in the Front Page editor: as I finished, Dirtlink broke my connection to the internet. Now Front Page says it cannot SAVE MY WORK even though I have reconnected. I have managed to cut and paste what I wrote to Word, and now I can dump Front Page, then reconnect, then cut and paste back; but this is a colossal waste of time and a hideous kludge, and a very bad defect in Front Page. If anyone knows a fix to this I would be grateful.

A program that needs to be connected to the net in order to SAVE on a LOCAL DRIVE was designed by a fiend who deserves to be outlawed.


Sunday, August 2, 1998

The site is getting a lot more hits, and some publicity out there, I guess -- and suddenly the spam. Is HOTMAIL ever legitimate? That is, do any legitimate users have hotmail as a return? I am not sure I recall any but I have not examined my mail list. I do know that I have got about 30 unwanted messages telling me about products and services I don't want of need, and nearly all had hotmail as return address.

I'm wondering what machine to put W 98 on. I have Pentafluge, which started as a Pentium 25 or 30 -- whatever the slowest Pentium was -- and which was a test site for Windows 95 when he was one of the very first Pentium machines. He now has an Intel Overdrive and in about 5 minutes will have a Kingston 200. He also has an IDE DPT SCSI and an IDE Creative sound board. Pentafluge has been used as Niven's work station and occupies an important bit of real estate; I'd as soon have a powerful machine there, but it has to WORK.

Two possibilities for that part of the Great Hall: Winnie, who now seems quite stable at 66/200 with the Winchip, running W 95 b because the Ricoh Media Master does not work with W 98, and for now I want that capability although I may give it up later; and Pentafluge upgraded with the Evergreen and running Windows 98. I confess I rather like the notion that the same old fussbudget machine that was once one of the fastest machines on Earth -- it was a VERY early Pentium chip obtained from a VP at Intel and delivered insured for some absurd amount -- and used as a beta site for W 95 has moved from Windows 3.11 to Windows 98. Not many can have made that Odyssey.

A third possibility is that I build a Win 98 machine from scratch. One of these days I want a P-II machine, although I am in no hurry for that. I could build a system more or less identical to Winnie, Winchip 66/200, but with Windows 98 rather than 95b, and in fact I probably will build such a machine.

Pentafluge with that ISA SCSI is never going to be a screamer. Has an ISA thin coax 10 megabit Ethernet too. Might be interesting to see if he is much improved by Win 98; if not I can reinstall W 95. Have to think on this.


birdline.gif (1428 bytes)