Microsoft's Largest Piece of Software Weighed More Than 40 Pounds (pcmag.com) 82
joshuark shares a report from PCMag: The official Windows developer documentation team at Microsoft decided to ask Microsoft Archivist Amy Stevenson "What was the largest piece of software we ever shipped?" The answer may surprise you... [T]he award goes to Microsoft C/C++ compiler with the Windows SDK, which was released in 1992 and weighed over 40 pounds. It included Microsoft C/C++ 7.0 in a box that was more than two feet long and allowed a developer to produce MS-DOS, Windows, and OS/2 applications. As Stevenson points out, "we never did that again," and the next product launched was Visual C++.
RTFM (Score:4, Funny)
Re: (Score:2)
Ah yes, the orange wall. Later replaced with the gray wall.
Re: Meh (Score:2)
I'm wondering if the coders are also writing the documentation. That would surely lead to burn out as well as it's hard to switch back and forth from the "master programmer" and "average Joe" mindsets.
Re: RTFM (Score:2)
I kept one of those on my shelf as a curiosity until just a couple years ago. I used to work in a university, and a prof gave it to me after a Microsoft rep had donated it to her. I was astounded at its heft!
As to how context sensitive help worked back then? Yes, this!! Often the help tool these days is no better than a shortcut to Bing. Just a poke in the eye . . .
Re: (Score:2)
Yes, the Microsoft docs broke down quickly I think. Visual C++ was good I felt, especially with contextual help, but with Visual Studio they replaced all the help files with html documents that were badly written and mousing over functions stopped giving useful contextual help.
Re: RTFM (Score:1)
Better Help Facility (Score:5, Informative)
The context-sensitive help on some of those older compilers was better than what we have today.
QuickBasic in particular had a wonderful feature where pressing F1 would pull up the relevant help information. Then Microsoft went through a long period in which pressing F1 was likely to pull up help for a different platform (MFC, Windows, other languages) than what was needed. The C compilers were particularly bad, as MFC function calls had identical names to the underlying Win16 Windows calls. This carried over to OS/2, Win32, and .NET. It seemed that any help request involved wading through a billion irrelevant variations on the same call. The OpenFile, OpenFileA and OpenFileW thing didn't help either.
The incredible thing was that the ancient help facilities were off-line. They worked great without access to the Internet. This can be a big deal in a secure development environment, or in a factory, or in a remote location, or whenever developing without Internet access.
Re:Better Help Facility (Score:5, Insightful)
The incredible thing was that the ancient help facilities were off-line.
Yet we called it "online help", which had an entirely different meaning at the time.
Re: (Score:1)
Then what was "off-line help" at that time?
Re:Better Help Facility (Score:4, Insightful)
A large book
Re: Better Help Facility (Score:2)
On-line used to mean a system that was up and running and accessable; "the dot matrix printer is on line". Periphials, mostly printers had a button and an indicator marked "on line" indicating that the parallel cable connection to the PC was now active and ready to be used.
"On line" help had the same meaning, that the documentation was on the disk and accessable through the program's interface.
When the internet and AOL became used by the general public, this term has become very muddled, and "on lin
Re: Better Help Facility (Score:1)
Re: (Score:2)
Ah yes, the golden days of VC++ and the tight integration into the MSDN docs. These days it's Russian Roulette as to whether you'll get a meaningful answer when you hit F1 in the code
Going back even further, I remember reporting an issue to MS about QuickC. Somehow the conversation got around to the fact that nobody at MS used QuickC for "real" development. They used the "big compiler" along with their homegrown ED editor. I was flabbergasted that they didn't see what a productivity win it was to be abl
Re: (Score:3)
QuickC was fast, and it highlighted the line on which an error occurred.
Today, we have three choices:
a) Use VisualStudio / Visual C++, which will be slow, but will highlight the line on which an error occurs.
b) Use VIM, which is fast. It is fast to open, and quick to type ":1000" to get to the relevant line number.
c) Use a language like Python which associates an error with a multiple line numbers that sometimes have nothing to do with the line that needs changing.
I have often wondered if we could add
Re: Better Help Facility (Score:1)
Using Makefile's, then :make will help you with the second.. It jumps to wherever the error was..
Re: (Score:1)
Look at geany. It's a lot zippier than Visual Studio.
Vim is better than anything else for macros and search-and-replace, though.
Re: (Score:2)
Re: (Score:2)
It really is amazing how far down hill things have gone.
One of the reason to chose Microsoft in the early 90s was how much better and more complete their documentation was compared with everything else. I would say that held thru to the early 2000s. After that it all got dumbed down, made less searchable and generally written to try and push people toward the technical decisions and technical choices the Marketing Department had decided were 'the future' rather than to simply describe the system.
Re: (Score:2)
One of the reason to chose Microsoft in the early 90s was how much better and more complete their documentation was compared with everything else.
I dunno. MS's documentation was quite good, but Borland C++ up through about 4.5 also came with documentation that took up about three feet of space on a bookshelf.
Re: (Score:1)
Re: (Score:2)
I also remember Borland shipping on a stack of dozens of diskettes. It was pretty impressive.
Re: (Score:1)
Re: (Score:2)
I asked some long time Microsoft employees about what happened to the quality of their documentation, and what I was told is that there was a distinct documentation group and the number of technical writers used to be significant, and you had people making the documentation that actually SWEs and/or had some knowledge and understanding of what they were writing about, and they would make sure sections were complete, and items cross referenced, etc.
But they were seen by executives as a big expense that could
Re: (Score:2)
Re: (Score:2)
To this day Microsoft's C compiler produces crappy errors and warnings. GCC is far superior, in that the information it gives you is usually helpful in fixing the problem.
Re: (Score:2)
Microsoft is pretty good about including usage examples too, in a time long before stackoverflow that was really useful.
Re: (Score:3)
Even today it's junk. I used to like Excel, and now it's an incomprehensible mess of too many features, but the help system is absolutely atrocious. Everything in Office is dysfunctional with help, and you're better off just searching the web for third party info.
Handy Manual [Re:Documentation] (Score:3, Funny)
It's amusing that the phrase "handy Manual" means different things in different languages.
In English, it means that the user's guide is close at hand and useful.
In German, it means that the user's guide is on your mobile phone.
In Spanish, it means that Manuel is the handyman.
Re:Documentation (Score:4, Interesting)
Two different early 1980s memories:
A whole row of manuals about the PDP-11/34a and RSTS/E, though the one that documented all the "syscalls" (OS APIs) was by far the most popular one, often not on the shelf so I'd have to track down who had it.
And then just a couple years later I was on Unix, and was shocked, surprised, and overjoyed by the amazing command "man" (and the man -k alias "apropos"). Books, once among the most valuable of things, were already headed to obsolescence.
Borland C++ was almost as big (Score:5, Informative)
Circa 1992, I bought a copy of Borland C++. The UPS guy needed a hand truck -- dedicated to just my copy's boxes -- to get it into my dorm. From what I recall, it had approx. 5 or 6 linear spine-feet of books and filled two entire shelves. And it was on CD... a friend got the floppy edition, which had a THIRD box containing at least 40 or 50 floppy disks.
By comparison, GFA Basic 3.0 for Amiga came in a 2-inch binder with 2 or 3 disks. HiSoft DevPac Amiga came in a 1-inch binder with 1 or 2 disks. Compared to Amiga stuff, PC software was SHOCKINGLY huge. I think WP4.2 for DOS came on at least 8-12 5-1/4" disks, compared to 2 or 3 for the Amiga version. WordPerfect for Windows was *scandalously* huge, weighing in at ~35 disks (more than Windows 95 *itself* did).
I think the grand prize went to Novell Netware... something like 80+ disks in the box, plus another 30-50 patch disks that had to be downloaded & unzipped to formatted disks... one... by... one. It literally took 2 days just to *prepare* to install it.
Re: (Score:3)
I got the floppy edition in 1993.
No idea how big the package was in cm or feet, but it was huge.
Re: (Score:2)
And it was a long day of work popping in those floppies to load it into your computer. An ordinary 8 hour day was not enough time.
Re: (Score:1)
Re: (Score:1)
Insert Disk 1 (again)
Re: (Score:2)
Not sure.
But I installed Linux Slackware (some 0.8 or 0.9 version) same year. Also minimum 20 disks. That took from 23:00 (11:00 PM) to roughly 4:00 (AM) in the morning.
Re: Borland C++ was almost as big (Score:1)
Re: (Score:3)
Amiga development vendors assumed you already had a copy of all three volumes of the ROM Kernel Manual, and Mac vendors assumed you had "Inside Macintosh". The RKM contained everything I ever needed, with great C and assembly examples in the Intuition and Exec volumes. Having said that, the Amiga's OS wasn't nearly as complex as Windows/Linux is nowadays, and there just wasn't nearly as much to document.
Re: (Score:2)
THINK Pascal and THINK C for Mac still came with two fat manuals each: the language/compiler/IDE manual specific to the language, and the THINK Class Library reference (application development framework) that was the same for both. But the Microsoft Platform SDK at the time came with the Windows equivalent of Inside Macintosh.
Re: (Score:2)
Think C was what hooked me up into C++ and Application Frameworks (and Design Patterns).
Only 2 manuals, yes. But everything was inside. Probably for a long time the simples way to program on a Mac.
Wow, I think I still have it in my basement.
Re: (Score:2)
I still have my THINK Pascal and THINK Class Library manuals. They're very well-written and typeset. They remind me of a better time when people cared about documentation.
Re: (Score:2)
Maybe for a desktop, but that's nothing compared to the complete documentation for VAX VMS. It came in a set of dark orange binders, so that they could be updated easily and was commonly refereed to as "the orange wall."
Not quite true... (Score:1)
As someone who bought it, the C/C++ 7.0 that was sold at retail was large but it didn't come with the SDK. I had to order it separately and it was just as large and just as expensive.
Re: (Score:2)
If you look at the box shown in the video it says "& Windows Software Development Kit", so clearly this version included it.
Re: (Score:2)
Oracle (Score:2)
I have a copy of Oracle for VMS that must weigh at least 20 pounds. The box is only a foot wide, but it contains the software on both 5.25" floppy and TK tape cartridges, as well as a dozen or so thick manuals. The only other box I've seen nearly as large is a 90's copy of Alias/Wavefront - or whatever their 3d product was called for SGI machines.
The before times (Score:2)
Re: (Score:2)
Re: (Score:1)
And before expertsexchange.
What does Expert Sex Change -- is that where you go if you want to be an Apple developer?
Re: (Score:1)
Re: (Score:2)
Re: (Score:2)
Before UseNet/UUCP we had BBSs to contain our flame wars. FidoNet, anyone?
I still have both the dark blue and the puke green Lattice C compiler manuals somewhere. I doubt the floppy disks are any good though. I also have selected parts of the last version of the Tru64 docset in my office bookcase, taking about 3 feet of linear shelf space. The DEC 8400 that it came with is long gone.
I carried it home on a bike (Score:3)
Re: (Score:1)
Re: (Score:2)
I purchased a copy of the Windows SDK paper documentation at a discount book store - it was brand new in shrinkwrap and I got it for fairly cheap when I found it. But it was a massive box of manuals in a slip case. Still useful reference books on the Windows API.
I had that, it was _very_ expensive! (Score:2)
Just the OS/2 beta with compilers & docs cost something like 3000 NOK, so about $300 at the current rate.
Terje
Simple fix (Score:1)
-Osize
not true (Score:2)
It's not punch cards but pretty sure this is more than 40 pounds...
https://www.smithsonianmag.com... [smithsonianmag.com]
Re: not true (Score:2)
Re: (Score:2)
So not only did you not read TFA, or even TFS, you didn't even read the HEADLINE.
Re: (Score:2)
Re: (Score:2)
The article, summary, and even headline make it clear that it is about the largest 'software' ever sold by MICROSOFT. Your post claims that the article is 'not true', then goes on to list things that have absolutely nothing to do with Microsoft products.
Re: (Score:2)
Re: (Score:2)
that's not a manual... (Score:2)
... (looking at VMS), now that's a manual.
Good old VMS with the wall of orange manuals, or grey for the later versions.
The paper was gorgeous (Score:2)
Re: (Score:2)
If you used this every day, you ended up referring to these documents a lot. Good quality, durable paper was important!
I worked for a company that was part software house, part ISP and the software development team had at least one of these boxes on a shelf. It was hefty, but so was the Solaris manual set.
The Solaris manuals didn't come in one box, but they occupied as much or more shelf space once you had the manuals lined up for the OS, OpenWindows [1], system libraries, compiler manuals, additional netwo
Hmm (Score:2)
I'll have to weigh the O'Reilly XWindows manuals up on my top shelf. I fesr taking them down, as a few pounds of that includes years of accumulated dust.
IBM manuals (Score:1)
Back in the day when IBM used to ship full sets of paper manuals, I worked in the CICS team (FWIW, CICS is an "online transaction processing system" dating back to around 1968, and it still runs a load of the critical banking and insurance backend systems today).
I'd estimate that for each release of CICS we'd get somewhere around 60 pounds** of manuals. I think the last time we got paper manuals was around twenty years ago.
The MVS (now z/OS) team next door to us would get somewhere between 5-10 times as man
The days of the tray of floppies (Score:2)
I ran the support center of a retail store in the 90s. The failure rate of one 3.5" disc in a raft of 20 was off the charts. We kept a set of office discs on hand to replicate the bad ones on demand. I was ecstatic when distributions on CD became a thing.
Visual C++ 1.0 (Score:1)
My first C/C++ Compiler was MS Visual C++ 1.0. IIRC it came on 14 3.5" floppies. There were three books. The thickest described the Microsoft Foundation Classes. The one describing object oriented programming was actually a good introduction to C++. Classes, inheritance, and operator overloading were the only features that set C++ and C apart back then. There were no templates.
I was sad find out that the compiler was unable to produce DOS binaries. All it could do were 16 bit Windows applications. There was
*Laughs in O'Reilly* (Score:1)
Yeah software typically "weighed" more (Score:2)
Because back in those days you had to assume the user didn't have so much as a BBS they could dial into to get the help they needed, not to mention connect charges, so there was the need for thick manuals that covers every feature and use case. Complex software had to be distributed on multiple floppy disks (could be as much as 20 or more) because most people didn't have a CD-ROM drive. Not all of the disks would be used by the user, but they had to include every library and feature set because they are cat
Re: (Score:2)
But if you were to print out all the documentation for current software... for equivalent programs that would consume a tiny fraction as much paper. They just don't document like they used to. Do we still need them to? I don't know, but I've seen times when it would be nice.
I have that version on my shelf, (Score:2)
beside the MS PDK,that includes Pascal, C, and Assembler. The PWB Programming environment that SW used is the best I've ever used. It had a meta-language that you could use across multiple files and languages. :)
One of my early projects was Life. Seeing 'golly' made me realize what a good programmer could do, lol.All that documentation is about 6 feet of shelf space.
It was so big it felt impressive jus 2 carry 2 car (Score:1)
Pounds = large software? (Score:2)
The blurb says the c++ compiler was the "largest" software ever released, going on to quote it in pounds. But is weight really the right criteria for measuring the largest software? I hereby release a special version of pwd with documentation on tablets of stone. I win!
Short memories at MS (Score:3)
"[Microsoft C/C++ 7.0] allowed a developer to produce MS-DOS, Windows, and OS/2 applications", really? A Microsoft product from 1992 supporting OS/2? No way in hell, LMAO.
Well, I happen to actually have this product and it doesn't support building OS/2 applications at all. (the previous version, 6.0, did, though). Not without some serious hacking and borrowing code and binaries from other products, anyway.
Knowledgeware CASE tool (Score:1)
Knowledgeware used to ship their Information Engineering Workbench CASE tool in a velvet-lined box.
Those velvet-lined boxes have disappeared. Even ebay doesn't seem to have them. That was back in the days of 5.25" floppies. People thought CASE tools were going to be hot stuff.