A little while ago, I mentioned that Windows 95 had some defenses against installers that overwrite a file with an older version. These defenses kicked in whenever it detected that a setup program had finished. But how does it know that a setup program was running?
It guesses.
If the name of the program contains the word “setup”, “install”, or some other magic words, then it is considered to be a setup program. Here’s the list of magic words.
| Magic word | Notes |
|---|---|
| setup | |
| install | Redundant |
| inst | |
| imposta | Italian? |
| ayarla | Turkish? |
| felrak | Hungarian? |
The entry for install is redundant with inst, because anything that contains “install” will also contain “inst”. My guess is that “install” came first, and then later somebody found that a lot of setup programs were called “blahinst” for various values of “blah” so they added an entry for “inst”, but failed to remove the redundant entry for “install”.
If there are no matches on the program name, Windows 95 also checks whether the path to the executable contains the word “Setup”.
In the above two cases, the file check is delayed until the next start, because some setup programs will realize that the file is in use and cannot be replaced, so they use ExitÂWindowsÂExec to exit Windows back to MS-DOS, run a batch file, and then start Windows back up. We have to wait until the restart of Windows to catch any files that were improperly modified by the batch file.
As a special bonus, Windows 95 does a live file check after any multimedia driver installs via an INF file. I guess the multimedia team discovered that a lot of drivers overwrite system DLLs in their INF files, so they asked for a cleanup pass afterward.
Are you seriously trying to mansplain Windows API to a MS Principal looking at the source code?
Half of the nonsense you just spewed doesn’t even apply to 95. NT kernel? It didn’t exist back then mate. Stay in your lane. Lay off the AI.
Ok, so instant.exe is a setup program…
That may be true, Mr Chen, but I don’t recall any of Microsoft updates in those days BRICKING an end user’s system like they do now.
Nice try at the damage control, though.
And… what does this have to do with the article’s topic? Raymond is clearly speaking about third party installers, not system updates (which didn’t even exist in the times of Windows 95).
Enrich, something tells me you never actually used Windows 95.