arlolra / tor-browser-bundle Public

 * Bump webrtc to branch-heads/58

 * Also, bumps go-webrtc and snowflake to the latest.

 * Removes the need to re-run ninja for FakeAudioCaptureModule since
   "pc_test_utils" includes it.

 * Update to the new default bridge

There is a prebuilt gn binary for linux64, but none for linux32, and
anyway we prefer to build from source for reproducibility reasons.

Version bumps, Changelog update, and update of config.yml

Adjust the comments to account for recent Torbutton changes.

…ubridge02 bridge.

https://lists.torproject.org/pipermail/tor-project/2017-January/000906.html

Thanks to arma for finding that one

Allow full read access to all files under /usr/lib.
Allow full read access to /Library/Preferences/com.apple.ViewBridge.plist.
Allow writes to TorBrowser-Data/Browser/profiles.ini (otherwise, a new
  browser profile is created each time the browser is opened).

 * Recent upstream changes require manual intervention when doing the
   initial fetch, which we'd like to avoid.  For now, just sync as far
   as we need.

 * Snowflake depends on go-webrtc, which in turn depends on libwebrtc,
   which adds significantly to the build.  webrtc is special, having its
   own build system that brings in lots of Chromium dependencies.

 * Further, building cgo reproducibly requires go 1.7.3 and a compiler
   that supports the following flags,

     -gno-record-gcc-switches
     -fdebug-prefix-map=$WORK=/tmp/go-build

- Don't ignore errors (except unzip exit status 1 or 2)
- Quote $@ and $1
- Work with absolute filenames and filenames starting with a dash
- Pass many files per chmod invocation (much faster)
- Pass $UNZIPOPTS, like $ZIPOPTS
- Reuse dzip.sh in re-dzip.sh

The (re)generated zip files are identical.

After trying to apply the fix for #20147 it turns out that we had an
error in the Firefox descriptor for Windows: the webapprt/omni.ja file
does not exist. There is therefore no need to re-zip it in order to get
a deterministic layout.

The current script we use ignores this error while the new one proposed
in #20147 is not so lenient anymore.