It may be true that people are partially looking back in rose-tinted glasses, but there's more than just an inkling of truth to their side. Casey Muratori (game developer for the Witness) has a really good rant [1] about bloat in Visual Studio specifically, where he demonstrates load times & the debugger UI updating today vs on an Pentium 4 running XP. Whether or not you attribute the performance difference to new features in Win10/VS, it's worth considering the fact that workflows are still being impacted so significantly on modern hardware. We were able to extract 100s of times more out of hardware and gave it up for ???
The Visual Studio 6 on Pentium 4 demonstration starts around 36th minute.
I used Visual Studio 6 for years, and yes, I can confirm, it was really that fast.
It's also not true that there were problems with more applications running etc as "Decade" claims. Or to be more precise, there were no problems if one used Windows NT, and I've used NT 3.51, NT 4 and 2000 for development, starting with Windows development even before they were available. And before that, Windows 3.x was indeed less stable, but it is the time before 1995. Note that the first useful web browser was made in 1993, internet as we know it today practically didn't exist. There were networks, but not web.
Maybe it’s possible for opposite things to be true if they happen to different people. I wasn’t a developer back then.
Windows NT required a multiple more RAM to run than the consumer versions of Windows (oh no, bloat!), and was much more picky about what hardware it ran on. Starting with XP, the professional and consumer versions of Windows have merged. We are so lucky.
> Windows NT required a multiple more RAM to run than the consumer versions of Windows (oh no, bloat!), and was much more picky about what hardware it ran on.
Allow me to claim that that is also not true, in the form you state it. Again, I've lived through all this, and I can tell you what that was about. The "pickiness" of NT was even at these times not about the motherboards and the chipsets. It was about the hardware consumer devices. Many things that probably don't even exist as the products today, like a black-and-white hand-scanner that scanned as you moved your hand over the paper and had only Windows 3.x drivers on the floppy with it. There was never a problem of having a developer machine running NT in any reasonable price range, with a reasonable graphic card, monitor, keyboard and mouse. And, at the start, a phone line modem transmitting some kilobytes per second!
The RAM needs did exist, but again not such as they are made to be believed by later distortions. If I remember correctly (it changed relatively fast), at the time NT was published, Microsoft had to deliver it claiming that it will run on 4 MB, the OS and the programs and the graphics, all had to fit. Let me repeat, 4 MB. It run, but not comfortably for bigger programs. But the point is -- as soon as you at that time had 8 MB you haven't had a problem. A little later, for comfortable work, 16 MB were more than a good choice. It was a hundred, two or three of $ more than the cheapest possible offer (yes, that were the prices then), but that was it. RAM was the only thing you had to care about to have NT running.
The point is, at that time there were enough those who didn't want to use Windows NT at all, clutching to 3.x and then 95 and these are those who promoted the horror stories about OS problems. But it was just their ignorance. 95 was also reasonably stable, unless you used, like many, some "utility" programs that were more malware than of real use (the "cleaning", "protection" or even "ram expander" snake oils were used by some even then -- no to mention that a lot of people believed they had to try any program that happens to access them).
The good development tools were good and stable, especially command line (in GUI areas, there were some snake oils among them too). But Word did crash even under NT, and even during the first half of 2000-s decade, and that's completely different story, that was intentional at that time for these products.
The word “reasonable” is doing a lot of work, here.
Most Pentium II systems were not running Windows NT. They were running Windows 95 or 98, which had arbitrarily severe limitations and lacked memory protections.
So, while it was technically possible to run lots of applications simultaneously on 256 MB of RAM, for most people it was a fun adventure in whether some buggy program has destabilized the system into needing to reboot to run properly again. Or whether it’s still usable with degraded functionality. In my case, that’s without using the cleaning, protection, RAM expander programs.
And even on professional operating systems, web browsers crashed a lot, and any other program that had to deal with untrusted input, which is basically anything that can open files or connect to the network, has gradually bloated as they learn security or add features.
> Most Pentium II systems were not running Windows NT. They were running Windows 95 or 98
Once again: only somebody using a computer not selected for serious development used Windows 95 and 98. No developer who knew what he was doing was using Windows 95 and 98 as his primary development machine. So if you complain about that, you used the wrong tool for your work. Like I've said, it was easy to install Windows NT, and I don't know any computer which wasn't able to run it, if it had reasonably enough RAM.
> on 256 MB of RAM
To illustrate "reasonably" once again, that changed at these times: I remember buying an AMD-based notebook in 2002 with 256 MB and running absolutely without problems Windows 2000 on it for a few years, before upgrading to 512 MB, which was the maximum for that notebook. And that was the time of Pentium III and IV, not Pentium II, and like I've said, I've run Windows NT on 8 MB computers, all with compilers, resource editors, debuggers and even IDE. And even before, I've run Windows 3.11 on 2 MB computer and used that for development too (the development tools being in text mode, of course).
> some buggy program has destabilized the system into needing to reboot to run properly again
Only on non-NT systems, and surely not developer tools. I used Windows 3.x and Windows 9x, and never had to reboot due to the developer tools "making system unstable." Not even on a 4 MB or a 16 MB machine.
> web browsers crashed a lot
I've used both Mosaic and Netscape, and before 2000 my main problem was surely not them crashing. Surfing mostly worked (only the pages loaded slowly, there were no CDNs then). Again, on a NT system.
I think we’re losing the plot. The ggp post was about doing all sorts of Internet programs at the same time on Pentium II era computers, and now you’re talking about developer tools on a Pentium 4.
Maybe it’s simultaneously true, that you could run many developer tools at the same time on Windows NT with hundreds of dollars of RAM, and attempting to run a bunch of consumer network programs at the same time (especially on consumer Windows) was asking for trouble.
I remember one of the attractions of IE 5 back in the day was how each newly launched window was its own process (not windows opened by the open link in new window menu option), so unlike Mosaic and Netscape, a crash in one copy of IE did not necessarily bring down all the other windows. Multiple windows being useful because surfing with a modem was slow regardless of CDN. Remember when Yahoo was scandalous, because banner ads took so much bandwidth?
> and now you’re talking about developer tools on a Pentium 4.
It's to illustrate that arguments are wrong: it's Decade who uses "256 MB" as an argument which is not "small memory" for Pentium II, and I illustrate that it was common in 2002 for notebooks, the time when Pentium IV was common for developer machines.
> The ggp post was about doing all sorts of Internet programs at the same time on Pentium II era computers
Let me check again:
"For example, IM, video/audio calls, and working with email shouldn't take hundreds of MB of RAM, a GHz-level many-core processor, and GBs of disk space. All of that was comfortably possible --- simultaneously --- with 256MB of RAM and a single-core 400MHz Pentium II."
OK. That is also obviously a bit off. 256 MB with Pentium II is quite a lot, as I showed 256 MB was normal even in 2002 for notebooks, as Pentium III was already common on notebooks and IV on desktops. Working with email -- at that time e-mail clients, if they used html at all, were limited to html formats of that time so "using email" completely worked, no crashes of system on NT (Outlook did have a limit of single PST having to be less than N GB, I remember that). IM too just worked, and also without crashes on NT.
That leaves "video/audio calls". Video calls were surely not common at that time, and I personally also haven't used audio calls.
But the "stability" problems you claim to have been common definitely didn't exist the way you claimed, as soon as one used NT, that is, since around 1994, or later on Windows 2000 or even later on XP or Server 2003, all NT-based. And as I've said, it was not that "too much" RAM was needed, as I've run NT on 8 MB with no problem.
So I still don't understand why you continue to stick to the narration that was simply not true. No, it was not that bad like you claim. Computers were quite stable even then for those who knew what they were doing. On NT, almost nothing crashed the system, except for failed hardware. Like I've said, it was that some apps were indeed less stable, like Word crashing or saving the invalid DOC file. But Excel, for example, while being in the same "suite" I don't remember to have ever crashed. I also don't remember browsers actually crashing, just the pages downloading very, very slowly.
The 256 MB number came from the ggp post. At the beginning of the Pentium II era, that was very expensive, but it was not the only issue with running multiple programs at the same time.
But clearly you want to have the last word, so I guess I should let you have it.
We gave it up for slightly higher profit margins enabled by hiring slightly less qualified programmers at a slightly lower rate.
In a similar vein, Industrial Light and Magic used to have a few highly talented people crafting incredibly intelligent solutions to make their movies possible: https://youtu.be/AtPA6nIBs5g
By now, most of those effects would instead be done using CGI and outsourced to Asia.
There's probably a long rant waiting to be written on this topic. Myself, I've observed how over the last four decades, CGI effects went from worthless, through novelty, through increasingly awesome, all the way to "cheapest garbage that can be made that looks convincing enough when the camera is moving very fast".
[1] https://www.youtube.com/watch?v=GC-0tCy4P1U