Reported by Marcin 'Icewall' Noga from Hispasec

Reported by Marcin 'Icewall' Noga from Hispasec

I wrote this as a quick hack after Gildas made the HTTP server bind to
all addresses instead of only loopback by default. It turns out it
lasted a bit longer :D

Lua was the only remaining user of ACL instead of (login/)password.
This is now fixed, so there is no need to clutter the httpd core.

This is way more flexible than ACL:
 - does not need admin rights to edit,
 - does not assume the clients have fixed IP addresses,
 - does not assume the clients are single user systems,
 - is a better understood concept than access control lists.

Ref #5814

macosx: make sure that VLC always shows a visibible video window when spawning a new vout (close #6418)

macosx: make sure that the subtitles menu item is correctly updated on playlist item change (close #6342)

Ref #6544

macosx: resize time field in the detached video window to make enough room for runtimes longer than 1 hour (close #6445)

Should close #6483

This fixes a bug whereby the configuration would not be saved because
Qt4 cleared the b_dirty flag instead of minding its own business.

The core determines whether the value of a configuration item as changed
using b_dirty. b_dirty is _not_ meant to signal changes in choices list.
Configuration callbacks blindly set b_dirty, so the UI can assume that
choices changed. Besides those callbacks should be removed as they do
not follow the locking model of the configuration and can cause crashes.

5a2390a7 is the first bad commit
commit 5a2390a7
Author: Jean-Baptiste Kempf <jb@videolan.org>
Date:   Sat Mar 31 22:20:58 2012 +0200

Close #6242

Close #6544

Close #6368

And close #6536

Use >= instead of > in range check because INT64_MAX could
be rounded up when converted to a double for comparison.
Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>

Close #6484
Signed-off-by: Rafaël Carré <funman@videolan.org>

Close #6118

VLC crashed when mounting multiple volumes at once

Remove -ansi flag for packages which don't build.

-ansi (and __STRICT_ANSI__) prevent the bionic headers to define (u)int64_t

Freetype exhibits a bug in gcc headers which typedef uint64_t u_int64_t

It seems like gcc, libebml, libmatroska, and taglib rely on standard headers
to define (u)int64_t which is the case for glibc even with -ansi.