2023-02-03

Index style

=> =========

Whenever you host a website or a Gemini capsule and your server

fetches an "index.html" or "index.gmi" file when a client asks for a

directory instead of a file, you have to ask yourself whether you want

to provide your documents as simple ".html" or ".gmi" files (with

meaningful file names) or as "index.html"/"index.gmi" files in

directories (which then have to carry names referring to the documents).

The former option is of course simpler, because you just have to write a

single document instead of having to create a directory first. (I've

also found a page once where the about page was explicitly linked to as

"about.html", as if this is also a matter of principle, but I can't find

the page any more.) This is also how it's usually done on Gopher, where

Gophermaps are usually really just index files in the true meaning of

the word. (It would also be way harder to write an entire article as a

Gophermap.) But there are also benefits to the latter option, which I

refer to as "index style".

One advantage is that you can leave out the file endings. If

your document is "example.com/article/index.html" instead of

"example.com/article.html", you can simply request

"example.com/article/" or even "example.com/article" to get it. So even

though the full path is longer, the path you would request is

significantly shorter. Of course you can also configure your server to

serve files even if the extension is missing in the request (I've even

seen ".html" extensions being stripped away from my requests (similar to

redirections, but I don't know if that's how it worked) on a site on

neocities.org), but just using "index style" works on most servers

without having to make additional configurations.

Another advantage is that you can assemble all files that belong

to a certain document in the directory that is "headed" by the index

file, and you don't have to make any adjustments if you don't have any

accompanying files yet, but decide to add some at a later point. This

applies to embedded image files in HTML documents for example, or

individual chapters of a larger document that you have split up. If you

have a Gemini capsule where you serve your documents in "index style",

and you want to mirror that capsule to the web, you can also convert all

"index.gmi" files to "index.html" files and keep both in the same

directory, for example. This way you can use the same root directory for

your Gemini server and your web server. (I've considered doing this

recently, actually.)

A third advantage is that you are less likely to end up with a

directory that doesn't have an index file (which might happen if you

would otherwise embed resources like images from a separate directory

without an index file, for example). This is mostly relevant for web

servers that don't generate index files automatically. But this also

makes it easier to "hide" automatically generated index files by

preventing their being generated in the first place (e.g. if want to

hide file modification times from users who don't know curl -I) if you

can't configre the server not to generate index files at all.

The big disadvantage, of course, is that *if* you want to have

automatically generated index files with file sizes and modification

times, then that won't work with "index style". (Since I'm thinking

about switching from Gemtext to plain text anyway, I'm also thinking

about using that functionality.)

EOF