Use the XComposite extension to get the composite overlay window,
instead of just using the normal root window. This ensures that
composited windows are covered.

This time, we hide it behind a config option because it might break
things more than it fixes.

Add command line option to lock TTY switching

Add build instructions

Makefile: include I3LOCK_VERSION in dist tarballs

fixes #129

Added ISSUE_TEMPLATE.md

Keep only the last attempt for retry_verification

Respect Xft.dpi for determining the unlock indicator’s scale factor

fixes #184

Fix memory leak when grabbing fails

return value corrected

This adds some feedback to a keypress which previously had no visible effect.

While the text “no input” isn’t the most descriptive, it was the only one I
could think of which fit into the unlock indicator circle. If you have a better
suggestion, let me know.

fixes #164

This was largely copied from the i3 configure.ac and Makefile.am.

Use RandR for learning about attached monitors

The code now tries the following options, in order:

• obtain monitors via RandR 1.5
• obtain outputs via RandR ≤ 1.4
• obtain screens via Xinerama

fixes #159
related to #101
related to #148
related to #149

When grabbing the pointer/keyboard fails, a new code path is activated, which:

1. Uses the standards-compliant _NET_ACTIVE_WINDOW root window property to
   determine the window to restore focus to.

2. Sets the input focus to the i3lock window, thereby possibly force-closing
   open context menus (works with e.g. Google Chrome, does not work with
   e.g. thunar, gedit).

3. Upon exiting, restores focus to the window from step ① by sending a
   _NET_ACTIVE_WINDOW ClientMessage to the root window. Note that this step
   requires https://github.com/i3/i3/pull/3027 in i3 to not mess up focus.

fixes https://github.com/i3/i3lock/issues/35

related to https://github.com/i3/i3lock/issues/35

Fix shadowing of getopt optind

Change the locale discovery procedure to treat empty string same as unset

This is explained in the commit message here:
f468f0b243

Only use -lpam when not on OpenBSD

Using 'optind' in getopt_long() shadows actual 'optind' being provided
by getopt, thus prevents any future options change from using optind.

See https://www.gnu.org/software/libc/manual/html_node/Getopt-Long-Options.html#index-getopt_005flong

This reverts commit 80d4452ec6.

fixes #130

Fix version

fixes #129

Restore intended behaviour and don't use mlock(2) on OpenBSD.

Add support for bsd_auth(3) as authentication backend

Also apply two security measures for OpenBSD:
- use explicit_bzero(3)
- mlock(2) works for non-root users too