5 ms·
Imagine a .JavaScriptObjectNotation instead of .json. Or .firefox-extension instead of .xpi There are only a few important extensions. just remember them. > C
by ffpip 6y ago
Imagine a .JavaScriptObjectNotation instead of .json. Or .firefox-extension instead of .xpi
There are only a few important extensions. just remember them.
> Computers are supposed to work for people, not the other way around.
Extensions are not for average computer users. They have icons and filenames. Windows also hides extensions by default.
- corty 6y agoThere actually was a short period where Windows 95 was new and had just introduced long filenames with long extensions to former DOS users. Then, for a short time, stuff like .SomeSoftwaresDocument was actually popular.
- chrismorgan 6y ago.JavaScriptObjectNotation would be terrible, because of its verbosity and because people call the language JSON: so .json is good. .xpi is a rather poor example: no one talks about XPInstall (and the majority of its surface area is now even obsolete), so .xpi is to most people completely meaningless unless they have encountered it before and know what it is. .firefox-extension would be a vast improvement over .xpi, because it says what it actually is.
- remexre 6y agoJSON is bad, but as someone who doesn't know what an XPI is offhand, .firefox-extension seems kinda nice?
- reaperducer 6y agoOr .firefox-extension instead of .xpi That's an excellent suggestion. WTF does ".xpi" even mean to someone who just wants to install a browser extension?
- glandium 6y agoWhat do ".zip" or ".exe" mean to someone who doesn't already know the answer?
- throwaway201103 6y ago> Extensions are not for average computer users. They have icons and filenames. Windows also hides extensions by default. As far as I recall, extensions became a thing with DOS. They actually had meaning to the OS, e.g. naming a file .exe would make it executable, as there was no other concept of file ownership or permissions. In Unix/Linux systems, filename extensions have always been for the user. Before GUIs and icons, they were a convention that let the user know something about the file contents. As far as the operating system is concerned, "." is just another character in the name and extensions are meaningless. Note that most binary executables don't have an extension at all, and other files have more than one (e.g. .tar.gz)
- reaperducer 6y agoAs far as I recall, extensions became a thing with DOS CP/M had extensions in 1974, five years before DOS. There may be older examples, but that's the first operating system I ever used.
- throwaway201103 6y agoMakes sense. I think DOS took a lot of inspiration from CP/M.
- wglb 6y agoAnd PDP-10 had them back in the 70s
- lokedhs 6y agoPDP-10 was an architecture, and there were many operating systems running on them. You may be thinking of TOPS-20, which indeed used filename.ext, where the extension was limited to 3 characters. There was also ITS, which was one of the famous operating systems on the PDP-10. On ITS, filenames consisted of two parts, rach being 6 characters separated with a space. So you could have a file named "abc def". The second half was often used in the same way we use extensions today, but executable files were named "ts name", where name was the name of the command. I'm not sure what format filenames had on TENEX and TOPS-10.
- 0df8dkdf 6y agoWell to be fare, file extensions are more of a Unix/PC thing. In the old day, MacOS Pre-OSX extension doesn't matter, extra info about a file are stored in another meta file. Extension are only made for backward/universal compatibility file compatibility on different platform. However I do agree with the parent post, computer ought to work for us, not the other way around. Then again changing convention is harder than just type 3 letter extension.
- schwartzworld 6y agoSeriously, I would understand not wanting to introduce new 3 letter extensions, but .txt is pretty ubiquitous. I'd feel weirder about seeing .text