Using SQLite's WAL (write-ahead log) can create less disk I/O while
still providing good consistency and durability.
This change uses SQLite's WAL with synchronous=NORMAL which can
cause loss of transactions on power failure. Only the transactions
which haven't been synced to the disk by the OS are lost. The
database itself will still be in a consistent state, but it might
not have all recent changes.
The class implements an API similar to OpenBSD's arc4random:
- A way to get a random uint32_t. (arc4random)
- A way to get a random uint32_t with an upper bound. (arc4random_uniform)
- A way to fill a buffer with random data. (arc4random_buf)
The current configuration uses a hacked up
win64-static-no-ltcg build environment until we
build our next set of real ones. As we do not want
to distribute our build env at this point the source
is obfuscated for now.
We utilize the appveyor cache to cache the compressed
build environment we downloaded. We do not cache the
uncompressed variant to save the time it takes to package
it up in case of cache invalidation. Cache invalidation
is keyed on changes to the appveyor.yml .
With the current setup, OpenBSD 6.0's g++4 on a VM with 4GB ram can't even
build Mumble, because the generated .cpp file from mumble_flags.qrc is
32MB.
This commit attempts to work around that by splitting our SVG flags into
separate resource files.
To make it easier to work with from the qmake build, we now also generate
a mumble_flags.pri file that will automatically add all necessary .qrc
files to RESOURCES.
This removes a lot of very hacky stuff. Its own job was to
let the manual plugin be a separate DLL.
Now that the manual plugin is built into Mumble itself, all
these dirty tricks can be removed.
Since we don't have to export a lot of symbols that we don't
use anymore, the linker can remove a lot of unused code for us:
mumble_app.dll before (1.3.0~969): 40.345 KB
mumble_app.dll after: 36.819 KB
Difference: -3.526 KB
As-is, the murmur.ini.system used by our PPA builds
(and Debian-based distros) are broken because the
script doesn't set the "logfile" and "pidfile"
options.
The email capture group accidently captured
the whole email part of the contact, including
angle brackets.
Update the capture group to only capture the
actual email address.
