| Commit message (Collapse) | Author |
|
|
|
|
|
|
|
|
|
|
|
They started sending links with "&username=june" tacked on after
the token. Don't know why since if you load the form with just the
token in the URL it still knows your username, and it's not submitted
as a hidden field or anything, but whatever.
|
|
20 is what everyone else seems to do. It's not too large putting
them all in right now, but it's infeasible to do forever.
|
|
What the fuck.
|
|
So that on start (and any time MAIL/MAILPATH change), any non-empty
mailboxes will be reported.
|
|
padvance_magic() returns -1 when there are no more paths left, not
zero.
|
|
This fixes "you have mail" showing right after checking and deleting
mail, resulting in a modified but empty mailbox. Also somehow fixes
"you have mail" always showing 3(!) times.
|
|
Eliminates the need for a shell inside the chroot, along with cat,
mktemp and rm. It's likely a bit faster too.
|
|
|
|
I'm not serving gopher anymore.
|
|
|
|
I haven't updated it in ages and I don't want to run (unmodified)
finger on OpenBSD since it likes to tell everyone your login IP??
|
|
Ideally these wouldn't require sh, cat, rm, mktemp.
|
|
cgit runs in a chroot. It is normal that it cannot look up user
info in /etc/passwd.
|
|
|
|
|
|
|
|
Migrating things over, I don't think it's worth continuing to serve
text over gopher.
|
|
Gotten a lot more used to running doas!
|
|
|
|
No idea why it made me do this...
|
|
|
|
|
|
OpenBSD col(1) doesn't understand UTF-8 and will delete all of it.
Stupid, especially when mandoc(1) tells you to use it to remove man
formatting. I shouldn't have to write something so trivial.
|
|
|
|
|
|
|
|
All this really does is make fixing problems more difficult.
|
|
|
|
This is a bit messy. pledge(2) calls based on [1].
[1]: https://kristaps.bsd.lv/kcgi/tutorial6.html
|
|
|
|
This reverts commit a64c3653d5df5e823fcdd07a540e240283816683.
Yeah this really doesn't work on OpenBSD. I kind of got into the
habit of explicitly running less anyway.
|
|
|
|
|
|
Use MSG_PEEK to determine where the request headers end and consume
only up to there, leaving the CGI process to read any request body
directly from the socket.
|
|
And only because kcgi expects it. This is what I wrote originally,
because every HTTP_ variable should be unset between each request,
but this doesn't track them.
|
|
|
|
Useless thing...
|
|
Oops! This whole time I thought OpenBSD xterm for some reason didn't
support focus events. It turns out allowMouseOps: false disables
them by default. Replace the disallowedMouseOps list with everything
but.
|
|
|
|
|
|
|
|
Surprisingly good for something written by a man like a hundred
years ago!
|
|
They were fun to make but I never actually used them.
|
|
I haven't used these in ages.
|
|
Otherwise upload won't actually work.
|