←back to thread

Why did Windows 95 setup use three operating systems?

(devblogs.microsoft.com)
304 points mooreds | 9 comments | 17 Nov 24 19:54 UTC | HN request time: 0s | source | bottom
Show context
dusted ◴[18 Nov 24 06:48 UTC] No.42170325[source]▶
Raymond Chens blog and book (The Old New Thing) is an absolute delight! I always had a big respect for how intuitive the Windows 95 GUI is, and reading his description of the thoughts and methods behind its inception, it's no surprise that it became so good. It seems like Microsoft was extremely pragmatic and reasonable in many of their endeavors back then. It's a wonder how it degenerated into the absolute unit of sh*t that is modern Windows (even if the filesystem and kernel is arguably a lot better, everything on top seems to be developed by an army of interns)
replies(5): >>42171191 #>>42171616 #>>42171979 #>>42172243 #>>42174602 #
andai ◴[18 Nov 24 11:50 UTC] No.42171616[source]▶
>filesystem is a lot better

In my tests it was 6-7x slower than on Linux (in VirtualBox on Windows). I assume by better you mean more features?

On a related note I used one of those system event monitor programs (I forget the name) and ran a 1 line Hello World C program, the result was windows doing hundreds of registry reads and writes before and after running my one line of code.

Granted it doesn't take much time but there's this recurring thing of "my computer being forced to do things I do not want it to do."

I also — and this is my favorite, or perhaps least favorite one — ran Windows XP inside VirtualBox (on Windows 10). When you press Win+E in XP, an Explorer window is shown to you. It is shown instantly, fully rendered, in the next video frame. There is no delay. Meanwhile on the host OS (10), there is about half a second of delay, at which point a window is drawn, but then you can enjoy a little old school powerpoint animation as you watch each UI control being painted one by one.

(Don't get me started on the start menu!)

Twenty years of progress!

replies(4): >>42171686 #>>42171830 #>>42172406 #>>42172567 #
1. jraph ◴[18 Nov 24 12:36 UTC] No.42171830[source]▶
> I assume by better you mean more features?

I would guess better on current Windows than on Windows 95. I don't know about faster, but NTFS is most probably more reliable than FAT32. And also more features of course, and fewer limitations. At least the file size limit (4 Gb) and ownership / rights metadata (ACL).

replies(2): >>42171926 #>>42177699 #
2. PaulHoule ◴[18 Nov 24 12:56 UTC] No.42171926[source]▶
Depends how you use it.

If you are handling a stupendous number of small files (say doing an npm build) then metadata operations are terribly expensive on Windows because it is expensive to look up security credentials.

Maven is not too different from npm in how it works except instead of installing 70,000 small files it installs 70 moderate sized JAR files that are really ZIP files that encase the little files. It works better in Windows than npm. Npm got popular and they had to face down the problem that people would try building the Linux kernel under WSL and get awful times.

Microsoft knows it has to win the hearts and minds of developers and they believe in JS, TS and Visual Studio code so they’ve tried all sorts of half-baked things to speed up file I/O for developers.

replies(1): >>42172197 #
3. akx ◴[18 Nov 24 13:41 UTC] No.42172197[source]▶
No filesystem is great in the lots-of-small-files case, partly simply due to syscall overhead.

There's a reason https://www.sqlite.org/fasterthanfs.html , SquashFS, etc. are a thing, or why even Europe's fastestest supercomputer's admins admonish against lots of small files. https://docs.lumi-supercomputer.eu/storage/#about-the-number...

replies(4): >>42172253 #>>42173942 #>>42176439 #>>42181557 #
4. red_admiral ◴[18 Nov 24 13:48 UTC] No.42172253{3}[source]▶
I guess that's where the "sqlite competes with fopen" part might help.
5. whoknw ◴[18 Nov 24 16:25 UTC] No.42173942{3}[source]▶
The reason LUMI is advising against many files is that it uses the Lustre parallel filesystem, which is notoriously bad with small files. See here: https://www.lanl.gov/projects/national-security-education-ce... .
6. mech422 ◴[18 Nov 24 20:16 UTC] No.42176439{3}[source]▶
That used to be ReiserFS's claim to fame - tail packing small files. Doesn't seem like any FS has really optimized small file handling since then.
replies(1): >>42177502 #
7. PaulHoule ◴[18 Nov 24 21:52 UTC] No.42177502{4}[source]▶
It is not so much the wasted space that bugs me as all the metadata and processing of that metadata.

For instance, the 70,000 files in your node_modules do not need separate owner and group fields (they all belong to the same user under normal conditions) and are all +rw for the files and +rwx for the directories. If you have access to one you have access to all and if your OS is access checking each file you are paying in terms of battery life, time, etc.

On Windows it is the same story except the access control system is much more complex and is slower. (I hope that Linux is not wasting a lot of time processing the POSIX ACLs you aren’t using!)

8. EasyMark ◴[18 Nov 24 22:14 UTC] No.42177699[source]▶
As someone who has done some fairly extensive reboot testing of “power failures” for fat32 and ntfs I can guarantee you that NTFS is far more robust but it’s not bullet proof
9. account42 ◴[19 Nov 24 09:35 UTC] No.42181557{3}[source]▶
SquashFS is a (read only) filesystem designed for small files.

Which shows that even if you don't want to call any filesystem great here, they differ vastly in just how bad they handle small files. Windows' filesystems (and more importantly it's virtual filesystem layer including filters) are on the extremely slow end of this spectrum.