Hacker News new | comments | ask | show | jobs | submit login
Doing Windows, Part 5: A Second Try (filfre.net)
244 points by nikbackm 7 months ago | hide | past | web | favorite | 59 comments

The previous posts of this serie have been discussed on HN:

* Doing Windows, Part 1: MS-DOS and Its Discontents (https://news.ycombinator.com/item?id=17403351) * Doing Windows, Part 2: From Interface Manager to Windows (https://news.ycombinator.com/item?id=17423724) * Doing Windows, Part 3: A Pair of Strike-Outs (https://news.ycombinator.com/item?id=17470844) * Doing Windows, Part 4: The Rapprochement (https://news.ycombinator.com/item?id=17523926)

There are a few blogs that HN should just scrape their RSS feeds and put on the front page, and this is one of that few

You weren't joking. Just skimmed through the rest of the blog and this is an absolute treasure!

Good place to start for others: https://www.filfre.net/sitemap/

Thanks a lot. From how in depth these are, I assumed your links went back for a year, instead of starting just 24 days ago!

(Why isn't there markdown to do unordered lists on HN?)

> Why isn't there markdown to do unordered lists on HN?

You can just leave a blank line between each list item.

* Doing Windows, Part 1: MS-DOS and Its Discontents (https://news.ycombinator.com/item?id=17403351)

* Doing Windows, Part 2: From Interface Manager to Windows (https://news.ycombinator.com/item?id=17423724)

* Doing Windows, Part 3: A Pair of Strike-Outs (https://news.ycombinator.com/item?id=17470844)

* Doing Windows, Part 4: The Rapprochement (https://news.ycombinator.com/item?id=17523926)

Excel installing a captive Windows, a technical detail they could have hid, was instead played up to demo the windowing GUI, in a better implementation of shareware than nearly anything that came after. This ingenious decision really ought to be more widely known and celebrated, but seems to have fallen out of the abbreviated narrative around Windows' (or Excel's) history, and it takes specialized sources like this one, or books by Excel gurus [1] to talk more about this fact.

[1] https://books.google.com/books?id=3Uz7CAAAQBAJ&pg=PA8&lpg=PA...

I used a few apps with the captive Windows runtime back in the day. It was highly frustrating because the app it was bundled with worked great, but if you wanted to do anything else (eg view other files) then Windows wouldn't let you because it wasn't the full product.

As a user the bundled Windows wasn't perceived as a benefit, since it consumed so much more hard disk space and had higher memory requirements etc. The driver situation was often dire too since established apps had established drivers, while early Windows had whatever few Microsoft had written. Microsoft was also obviously pushing the Windows agenda, after many delays and much competition as outlined in earlier articles in the series.

In case anyone wants to fix it, I also have a bug report from the one time I used Windows 1 as a kid. I used the paint program and went to save. A standard file dialog didn't exist as part of Windows until 3.1 (1992), so before then each app was on its own. Paint gave a single line edit control where you typed in the whole pathname (no browsing or similar). Back then DOS had a 64 byte pathname limit but the dialog let you keep entering characters, so I did. When I hit save, nothing happened except brief blips of the hard drive light. I had to reset the machine and found that this buffer overflow had resulted in most of the entries in the root directory being deleted! A quick disk edit tool and they were fixed (a flag is set on directory entries to say deleted so you just had to undo the flag providing you didn't create new files to overwrite these entries).

An intriguing product from the era was Desqview/X. You got a full blown X server and so could direct Unix programs to display on DOS. It could also export DOS programs to other X servers. If it had succeeded, then in this alternate reality apps would run on unix and PCs would be good for games and accessing those "real" apps. https://en.wikipedia.org/wiki/DESQview#DESQview/X

> I used a few apps with the captive Windows runtime back in the day. It was highly frustrating because the app it was bundled with worked great, but if you wanted to do anything else (eg view other files) then Windows wouldn't let you because it wasn't the full product.

In other words, working as intended. I mean, that's basically what you got with old DOS apps. You could spawn a new command.com from within the app if you were lucky, but that's not nearly the same as multitasking. You had a stack of programs running and had to completely exit the second one again to get back to the first.

So windows didn't really take anything away but gave you a pretty good GUI.

I don't recall if this is accurate, but an example would be wanting to make a note in a text file. Under Excel and captive Windows you would have to save and quit, exit Windows, start your note app under DOS, take the note and then repeat the sequence in reverse order (perhaps minutes).

Under a DOS app, you hit the hot key for your TSR[0], made the note and continued (a few seconds). As the series mentioned, being a power user under DOS was very productive. Lotus 1-2-3 especially had a notable menu/macro system, as did WordPerfect etc.

GUI environments are not very useful for one program. They have far higher system requirements, and perform slower than native text environments. And they especially do not play nice with others. This is why there was so much scepticism at the time amongst those paying money for hardware and software. Of course new customers and more of them were just on the horizon as the series will soon reach.

[0] https://en.wikipedia.org/wiki/Terminate_and_stay_resident_pr...

> GUI environments are not very useful for one program.

Yes they are, which is why lots of DOS apps used one-off GUI environments at the same time that Windows captive and runtime apps became an additional way to acheive the same thing without building your own GUI from scratch.

> And they especially do not play nice with others.

That's not a GUI issue, that may be a “captive version of a desktop environment” issue that affected the captive Windows environment; there's nothing inherent about one-app GUI environment that makes it not support any of the (hacky and unreliable, mostly, because DOS wasn't built as a multitasking environment) multitasking options that DOS apps worked with as well as non-GUI apps.

You could probably find this info in the trade press like InfoWorld, but yes, Microsoft was essentially giving-away Windows to get it catch on. 3rd party ISVs also used "runtime Windows", and you could get a full copy free with a MS mouse, or Visual Basic, or etcetera. You could also give copies to all your friends and coworkers, there was no copy-protection, and the software police didn't mind. Didn't really catch on until Windows 3 however.

I seem to recall reading that MS only started doing the activation thing in XP because the BSA threatened to throw them out.

Frankly MS has held that it is better, for them, that the public pirate MS software than switch to alternatives.

Also, private usage of MS software is seem by them as a sales pitch towards businesses. It allows MS to argue that potential employees already knows basic operations, so there is less required training for new hires.

> I seem to recall reading that MS only started doing the activation thing in XP because the BSA threatened to throw them out.

I don't buy it. Microsoft was so huge that I don't see why they would care about the BSA. It seems more likely that product activation really has helped Microsoft make more money.

Microsoft co-founded the BSA because the software piracy rent-a-cops of the day -- the SPA -- sided with the DoJ in the antitrust suit against Microsoft.

It didn't catch on until Windows 3 because Windows 3 was the first Windows version to come preinstalled on new PCs. Microsoft knew then what the Linux folks never understood -- preinstallation is key.

That preinstallation is key was widely understood, which is why Microsoft structured their agreements with OEMs to make selling machines with non-Microsoft operating systems very difficult.

I knew nothing of that Excel story before this article, despite being generally familiar with a lot of MS and Windows history. It definitely seems more obscure than it should be, at least to me.

(At the time I was still 100% MS-DOS; I never encountered Windows until 3.0. Plus I was young enough that my main exposure to business/spreadsheet software was through my dad, who was a loyal 1-2-3 user and remained so for many years afterwards.)

Two thoughts.

1. I can't get over how primitive the Windows UI was compared to Mac years after Mac shipped. Apple began the Macintosh project in late '79 and by Jan '84 shipped both the Mac and the OS. Microsoft began Windows development in '81 and didn't really catch up UI-wise till Windows 3 in 1990. Meanwhile, Apple would squander its head start, not shipping an OS as technically sound as NT (1993) till OS X in 2000.

2. I have yet to read a flattering portrait of Ballmer and here he almost kills Microsoft going all in on OS/2. Was Microsoft successful in spite of him?

From all the available recollections, Ballmer never had any valuable foresight.

I wonder why the board of directors left him as CEO after Gates’ retirement.

Also reportedly Ballmer was disagreeing with Gates about how important the market of smartphones and tablets is, with Gates wanting to invest heavily in it. Oops!

Microsoft made a ton of money with Ballmer as CEO. I imagine the board was pretty happy with him at least for most of that time.

How much of that revenue can be attributed to Balmer though?

As an outsider it seems to me that all Microsoft did during Ballmer was to milk the cash cows created during Bill Gates and at the same time destroy the good will of developers and the morale of their employees.

I’m assuming any big company acquires a certain inertia and will keep producing money for as long as there is demand and the supply can meet it. Just look at Apple.

So what did Ballmer do that can be praised?

I won't defend Ballmer, but the guy was a salesman, a really good salesman. Not a technologist.

Also Bill Gates was "chief software architect" during the Longhorn/Vista thing so he's hardly absolved of blame.

In defense of the Longhorn/Vista blunder, the problem is that Windows XP was really good (at least compared to previous Windows versions), some would say good enough for people to not upgrade to newer versions ... if you remember, this was the first Win NT version meant for consumers. By comparison previously Win 3.x, 95, 98, ME, and in between were really unstable and frankly unusable in certain scenarios, I know because I used them all.

I have a feeling that even now Windows XP is more popular than people think and that they aren't reporting the real number of Windows XP installs in the wild.

So they were competing against themselves, against a previous version that has been wildly successful, enough so that virtually nobody upgraded it. I mean, besides people capable of assembling and upgrading their PCs, component by component, I'm pretty sure that everybody bought into Windows 7 and later via new PCs. And PCs becoming good enough to keep around for years, I think the writing was on the wall that Windows's revenue will eventually go down.

That really echoes what I've been asking for years. If you get a CEO that walks in, does barely anything, will anyone even notice? Short of any media blunders or completely nuking a company, how can you really mess up a company during your tenure such that you get in and get out with your ridiculous compensation?

Sure . . . but they might have made more money with someone with better technical skills. Additionally, Ballmer had some dogma (Jack Welch style up-or-out review system, the "everything has to be Windows" nonsense [1]) that did a lot of damage.

We'll never know. But Satya seems to be doing very well indeed with a combination of better technical direction and removing Ballmer-era toxicity.

[1] On one project I worked on was an embedded system, and I joked that the only way that Ballmer would let it ship (if he'd known about it) would be to rename it "Microsoft Windows, Event-Loop-and-Some-Interrupt-Handlers Edition," whereupon it didn't matter what the heck the software looked like on the inside. As long as it said "Windows" somewhere on the outside, he'd be happy.

My guess, he was a money man appointed by a board filled with money men.

Was OS/2 tied to the PS/2 and MCA? In my first job out of school in the early 1990s, all the developers had OS/2 workstations. It was a good OS, stable, multitasked well, and from memory Microsoft didn't really equal it until Windows NT4 or maybe even 2000. From my perspective OS/2 didn't fail for technical reasons, it must have been IBM trying to recapture the PC as a proprietary platform when the market really wanted something that would run on any PC clone.

Not strictly, but the original plan was to have Microsoft market OS/2 2.0 to the clone makers. So when MS walked away, IBM didn't have a good story for 3rd parties. OS/2 2 had drivers for stuff with names like "IBM 3142" and you had to know whether that was the equivalent of an Adaptec SCSI adaptor or HP LaserJet printer. The association with PS2/MCA certainly muted OS/2's reception though.

That OS/2 2.0 debacle is one of my favorite topics BTW. It ended up in the end much worse than that.

I was quite young back then, but I started on os/2 and after about half a year we switched to nt4. I never used any windows before that, but nt felt much more mature. Unfortunately I don't recall the version of os/2 in use, but we mostly didn't upgrade the workstations when we switched and windows still felt "better", although I do remember some os/2 fans at our place complained about nt4 being a memory hog when the switch was announced. Iirc we had it running nicely on 8mb 486 even, and the visual basic 5 ide ran lightning fast.

Stability wise, my os/2 time was probably too short anyways, but I don't remember either os as unstable.

I ran OS/2 on a 486. One of the issues that held back adoption was it’s system requirements - you needed a minimum of 8MB of RAM for it to run smoothly, but most systems being sold at that time had 2-4MB.

> I have yet to read a flattering portrait of Ballmer

On HN six days ago:


Thanks, missed it. Added to my reading list.

> 2. I have yet to read a flattering portrait of Ballmer and here he almost kills Microsoft going all in on OS/2. Was Microsoft successful in spite of him?

I don't have a lot to back this up on, just years reading books on the subject, but I always got the impression that Gates/Ballmer in the early Microsoft days made a fascinating good cop/bad cop routine. I will continue to wonder how much of what was attributed to either Gates or Ballmer in those early days was simply their assigned good cop/bad cop role (with the weird twistiness to it that I think the roles swapped in internal communications versus external communications; Ballmer was the friend of IBM, the respectable salesman, and Gates the hot shot kid that wouldn't take no for an answer / Gates the founder that loved technology and took everything seriously but Ballmer was the business man trying to reign in freedom) rather than their actual thoughts on the matter.

I certainly got the impression that early Microsoft was successful because of that two-headed dynamic. Microsoft could pet IBM with one hand and stab them in the back with the other, and IBM would let them try that again mostly because they succumbed to that good-cop/bad-cop routine quite effectively.

Some of the mythos internally at Microsoft always seemed to suggest that the dynamic played similarly inside the halls, in building the early culture.

I wonder if in Ballmer's tenure after Gates' retirement the problem was simply that he never quite found that replacement good cop to keep the dynamic alive internally (despite externally didn't really need a bad cop anymore). Satya to all appearances is nothing but "good cop", externally and internally and maybe that's all Microsoft needs now to survive. Early Microsoft probably wouldn't have.

Anyway, just supposition on my part.

During those days, the Mac was hardly seen in the wild outside US.

For example in Portugal, we had a single importer, Interlog, located in Lisbon and Porto.

For us, the only competition to PCs were Amiga and Atari systems, but they were seen as gaming machines by most, because MS-DOS was where business was.

By business I mean as terminals to mainframes, UNIX, or simple Novel Netware.

I never liked the PS/1 and PS/2 things IBM did.

It seemed like an obvious attempt to try to snare customers into a closed system. I never knew however that IBM tried to snare in Microsoft too.

An interesting read for sure.

Just the PS/2.

The PS/1 arrived several years later and was an attempt by IBM to reclaim some of the ground lost to the emerging "PC clone" architecture. It was an ISA bus machine with IDE drives -- the opposite of the proprietary MCA stuff going on in the PS/2's of the era (which by this point was largely regarded as a business failure).

Interesting that Excel was developed first for the Mac. Excel also used p-code [1][2] for its cross-platform abilities:

"had a compiler which, back in the 1980s, generated pcode and could therefore run unmodified on Macintosh’s 68000 chip as well as Intel PCs."

It would be interesting to see the first version of excel on Mac, the windows version in the article looks very familiar.

[1] https://www.joelonsoftware.com/2001/10/14/in-defense-of-not-... [2] https://news.ycombinator.com/item?id=9254060

Ballmer could have sink Microsoft in 1986. Didn’t knew that but makes sense. How many long term bids are we missing in 2018 because Ballmer-like executives are now in power ?

I assume that some variant of "bundled Windows Runtime" is the reason why trial versions of various of various productivity SW that came with my first PC had MZ stubs, that didn't print "This program requires Microsoft Windows", but EXEC'd something like "%COMSPEC% /C WIN.COM PROGRAM.EXE"

At the time, Lotus was faced with a choice on where to go with Lotus 1-2-3: OS/2, Windows, or text mode. They backed the wrong horse. They were a big company at the time, could have easily bet on all three, and stayed in business.

As the years go by, and new stories like this come out, I find increasingly growing respect for Bill Gates.

I love how un-apologetically "skeumorphic" this blog is. Ah, I miss the days...

I miss that skeumorphic design. I can't wait until this flat fad goes away.

It's like somebody hit the computer UI design world with a stupid stick. And all the designers woke up without the ability to see in 3 dimensions, and lived a life as Flatlanders.

Gestures are fantastic though.

The best part is the loose threads. Taking the good with the bad from the real world.

How do you flag something as a dupe?

Just flag it, and make a comment pointing to the previous thread so other users and moderators can find it easily.

If there's no previous thread (i.e., a submission of that item in the past 12 months with a significant number of votes and of on-topic comments), it's not a dupe :)

It is not a dupe if it hasn't gotten tracktion before.

Guidlines: If a story has had significant attention in the last year or so, we kill reposts as duplicates. If not, a small number of reposts is ok.

Hadn't seen this article before. Enjoyed it. Good link and source for further reading.

Odd, posted this yesterday: https://news.ycombinator.com/item?id=17576304

HN is notoriously fickle. The same story can get 0 upvotes, then be submitted by someone else in a couple hours and then garner hundreds of points.

Anecdata: HN seems to prefer different content depending on the time of day - I tend to see more programming and software-related stuff in the late evening (PST), news in the morning/early afternoon, and general "interesting stuff" (e.g. Atlas Obscura) in the early evening.

Of course, this may well be confirmation bias, coincidence, or a matter of my browsing habits. It would be interesting to see an actual analysis of content discipline as a function of time.

It's a matter of who's available to upvote at specific times, and whether the topic makes for a nice read at those times. Sometimes a deeply technical topic or a very lengthy article aren't the best fit for the mood :). Other times they are.

How? When I post something and it happens to be recently posted, it automatically becomes an upvote. I'm generally fine with it, except when it has 1-5 votes and will never see the front page.

HN has so many secret behaviour that it became boring by now. I don't think I will bother posting a story again, because it's very hard to get some discussion to happen around a post (which is the whole point of posting a link for me), whereas in most Reddit subs, even if your post is in the -1 to 1 votes range, you can still get useful answers or interesting comments.

> How? When I post something and it happens to be recently posted, it automatically becomes an upvote.

It seems that sometimes, high karma accounts can submit a dupe. But I don't think it always works. Maybe it's a lower dupe submission check timeout.

As for submissions, I'd recommend trying to land in the American morning/early European afternoon. That's when most people seem to browse and thus you have a better chance of getting upvoted to the front page.

You may also be interested in this: https://github.com/minimaxir/hacker-news-undocumented

I'm in Turkey which is UTC+3, and I usually submit (well I don't usually submit, but) around noon or evening, which happen to be the moments when I usually check my feeds and my mail for newsletters. Strategical submitting seems too much to me, as if I was doing it for the karma points rather than the discussion.

> It seems that sometimes, high karma accounts can submit a dupe. But I don't think it always works. Maybe it's a lower dupe submission check timeout.

Yeah, yet another enigmatic "natural phenomenon" of HN. I appreciate the link you've given, but my actual point was that this sort of unknowable phenomena discourages participation. For example some time ago the [vouch] links disappeared for me. Some more time ago edits became impossible (2h is too short of a time, say I write something silly, 5h later I check back, and want to append an apology or a clarification; that's impossible, I need to reply to myself and hope people see that before scalding me). Also can not downvote after a day. What if I came across some misinformation that was older? Detaching threads is awful, because your words are removed from context; why not delete instead? Green usernames are an awful idea, most often I don't care about the poster, and why mark people because their accounts are new? Nobody needs that information. These sort of things prove to me that HN is not a "community", but more of an "order". We don't get to decide anything.

It used to be you could add something trivial to the URL like a trailing slash or some non-existent params et voila! New submission.

There is literally not a single comment addressing the content here at this moment :)

Yes, I'm not helping

I read that yesterday and I found it interesting as an history lesson and for the strategic insights, but I can't say I'm familiar with any of that. I didn't even know than windows before 3.1 was really used, and not just test versions

Applications are open for YC Summer 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact