Built my own "paste bin":
Most important features:
o Easy to use. Just do a "webdump filename" and you're done.
o Files expire. No risk of ending up with huge piles of junk.
o Everything is stored verbatim. No data mangling due to HTML forms
or stuff like that.
o When dumping multiple files in one go, they all end up in the
same directory. That directory provides an HTML index, but
there's no root index listing all dumps (on purpose).
o You can dump single files or whole directory trees.
o I'm not reimplementing stuff like authorization, transfer,
indexing, whatever. (Some of that would be inevitable when
building a modern web application.)
o The dump area is hosted on my own server. Plus, it's configurable
and easy to migrate to other servers. In fact, I do use webdump
on multiple servers.
o It's not anonymous.
o webdump does not work on shared webhosting servers. You need SSH
o There's no HTTP transport, that is, you can't use tools like
"curl" to dump stuff. You need the "full blown" webdump script.
Initially, I wanted to use unison to sync the files between the server
and the local machine. Turns out, that's almost impossible. Unison
only works with the same version on client and server. Arch has 2.48,
Ubuntu has 2.40. Gah, that sucks. I'd have to custom-compile Unison
2.40 on my Arch machines, I'd have to fiddle with config files a lot,
... Nah. Thus, I switched back to rsync.
Fun side project, related to webdump:
It's a simple script that generates random strings. However, those
strings look like "Laos-Fiepp-23", "Gaort-Loop-33", or "Kiemm-
Haott-18". That is, they are effable (for Germans, at least) and easy
to remember. Longer strings like "Xaost-Hok-Diamm-Joh-801" could even
be used as passwords.
webdump uses this script to create nice URLs, unlike pastebin.com
which simply spits out random alphanumeric strings.