3 ms·

I’m not fully convinced of my own position here, but I have long suspected that “zipping” at all is an abstraction that shouldn’t really be exposed to the user.

I’m not fully convinced of my own position here, but I have long suspected that “zipping” at all is an abstraction that shouldn’t really be exposed to the user. Like it’s a throwback to an earlier era. The basic trade off is size for speed, and it seems like most cases it could handled automatically at the level of the file system maybe. Like if I have a big file that I want to send, it sort of feels like the software I use to send it should do the work of preparing it for transport to the target system. Similarly, if there are files that aren’t being touched very often shouldn’t the file system figure out that file can be in “small, slow” mode? Or maybe if I know a particular file should be always fast or always small, I should have the option in the file properties for it to be Automatic, Always Small, or Always Fast, like a toggle right near the permissions controls. But not like a separate program that generates separate, smaller, slower files of a special type. Why can’t folders be automatically treated like singular archives when I try to do operations on them for which that makes sense? Why do I have to think about these details? I’m probably missing some important use cases and archiving features, but maybe those can be broken out from the default of “do the obvious thing automatically.”

mavhc 5y ago
Folders can be compressed archive files with the right OS/fake filesystem combination, although that's mostly ease of use. The main 2 reasons for an archive file are giving to someone else, or compression. The FS can do compression these days, so it's mostly about giving 1 file to someone else, not 1000. Using archive files to avoid excess file wastage seems like a bad idea, make the FS better instead
unfocussed_mike 5y ago
> Why can’t folders be automatically treated like singular archives when I try to do operations on them for which that makes sense? Why do I have to think about these details? Operating systems do blur the boundaries of this a bit, now -- you can explore zip files as if they were folders, mount disk images from files, and you can work with folders as bundles (e.g. a macOS application is a disguised bundle, an RTFD document is a directory with contents etc.) Some applications (like mail apps) have always done a good job of handling (combining, compressing) folders so they can be moved/sent as files. The problem is all of these application things are probably best done at a GUI level. At a filesystem/command line level you want the distinction to be explicit. Of course a few OSes ditch the distinction entirely and concern themselves only with objects -- OS/400 for example. Which might be closer to the world you are imagining.
- CRConrad 5y ago
  > Operating systems do blur the boundaries of this a bit, now -- you can explore zip files as if they were folders, mount disk images from files, and you can work with folders as bundles (e.g. a macOS application is a disguised bundle, an RTFD document is a directory with contents etc.) Also, AIUI, "Save entire page" in some browsers. You might think that should be a single .html file, but it's usually an .html file plus a directory of stuff linked in that file / on that Web pge, that the browser (and sometimes the OS) then does its best to pretend is just a single file.
teddyh 5y ago
You can write code to do anything, and therefore anything can be viewed as something else. However, all abstractions are leaky, so this is rarely usable in practice, since those leaks trip you up all the time.

mavhc 5y ago

Folders can be compressed archive files with the right OS/fake filesystem combination, although that's mostly ease of use. The main 2 reasons for an archive file are giving to someone else, or compression. The FS can do compression these days, so it's mostly about giving 1 file to someone else, not 1000. Using archive files to avoid excess file wastage seems like a bad idea, make the FS better instead

unfocussed_mike 5y ago

> Why can’t folders be automatically treated like singular archives when I try to do operations on them for which that makes sense? Why do I have to think about these details? Operating systems do blur the boundaries of this a bit, now -- you can explore zip files as if they were folders, mount disk images from files, and you can work with folders as bundles (e.g. a macOS application is a disguised bundle, an RTFD document is a directory with contents etc.) Some applications (like mail apps) have always done a good job of handling (combining, compressing) folders so they can be moved/sent as files. The problem is all of these application things are probably best done at a GUI level. At a filesystem/command line level you want the distinction to be explicit. Of course a few OSes ditch the distinction entirely and concern themselves only with objects -- OS/400 for example. Which might be closer to the world you are imagining.

CRConrad 5y ago
> Operating systems do blur the boundaries of this a bit, now -- you can explore zip files as if they were folders, mount disk images from files, and you can work with folders as bundles (e.g. a macOS application is a disguised bundle, an RTFD document is a directory with contents etc.) Also, AIUI, "Save entire page" in some browsers. You might think that should be a single .html file, but it's usually an .html file plus a directory of stuff linked in that file / on that Web pge, that the browser (and sometimes the OS) then does its best to pretend is just a single file.

teddyh 5y ago

You can write code to do anything, and therefore anything can be viewed as something else. However, all abstractions are leaky, so this is rarely usable in practice, since those leaks trip you up all the time.