Jo22 wrote on 2021-05-27, 09:10:
Today, in retrospective, Windows 98SE looks solid.
But back then.. Well. Stability heavily depended on hardware/drivers.
And XP was so much more stable, even before SP2.
That's about how I Feel about all the Windows O/S these days. I think it's because were not "daily driving" them anymore, and giving them a more dedicated role as a "gaming system" or a "retro system" vs. spending every day dialing-up to the internet and chatting online via AIM while reading 2-3 webpages till the whole thing gets an "illegal operation" or bluescreens - or in the case of Windows 3.1x - Stack Segmentation Faults, General Protection Faults, and other hex errors. Back in those days, I was reformatting my Win 98 SE machine at least once a year - starting over fresh, and hence I avoided 95, 98, and Me like the plague. I even preferred Windows 3.1 above them all because I could throw on Dr. Watson and 1mb Fort in the startup and basically never reformat or reload my 3.1 machine ever again.
Now I have Windows 95 running the same setup on 4 different NEC Versa Laptops, a 95 setup on an SSD that I've had for almost 4 years on my desktop, 98 SE just keeps running nonstop until I get tired of it and switch to a different HDD. Also, I think the newer productivity applications hobbyists are making for these legacy O/S are a lot more solid than the stuff we were using back then that likely had a far tighter production schedule and little to no actual passion behind them. Which means, tigheter, faster, better written code, because likely the author is using it his/herself.
That said, one of the other above mentioned drivers in 95 being Cantankerous - that is really true. Part of it is some companies still used driver installation methods similar to Windows 3.1x, while others chose to take the new path that exposes bugs in either patched or unpatched Windows 95. Prerequisites are not always clear for a certain program or feature - some will install on the original 95, while others need some kind of update that was released later that makes 95 more like 98 SE or even 2000 in some respects. In Windows 3.1 about all it did was wrote to the win.ini or system.ini, copied the driver files to C:\Windows\system\, and then restarted Windows 3.x and your hardware worked. 95 used the two methods, and by 98 SE we were mostly seeing it much like today - some kind of installer program just installs the files, sets up all the complex background stuff (system.ini, win.ini, protocol.ini, the binary registry values, plus any shortcuts and groups for the utilites bundled with) - and then you reboot and it all works.
Also, 95 was really the changeover from the "old-world" PC's with manual jumper configs on everything (ie everything 486DX2 on back with an ISA bus or janky early VESA local bus) that were designed in a world were MS/PC-DOS ran everything, to a world of PC's with USB, Plug'N'Play capability, PCI and some late ISA components that could take advantage of it. So it's little surprise a 386 DX40 might be happy running 95 while a Pentium from the same year as Windows 95 is a cantankerous beast. Actually, a lot of those "Windows 95 Certified" machines from 1995 were really older models that were shipped with Windows 3.11 and DOS 6.22 the before. Was researching the Samsung Sens recently and saw this in action on Google Books in an ad for the Sens 700/800 486/Pentium based laptops circa 1995 - they were proudly proclaiming "Now comes with Windows 95 Preinstalled" - on models that only a few months earlier were strictly shipping with MS-DOS 6.22 and Windows 3.11.
The benefit of 95 is it's lighter in weight on a base install compared to 98 SE, but 98 Se can come pretty close with some tuning, tweaking, and 98Lite. When my NEC Versa 40EC was running 98 SE, it had 98 Lite on there, ran pretty much as well as 95, but was far more stable.