| Commit message (Collapse) | Author | Age |
| ... | |
| | |
|
| |
|
|
| |
build-sunet-a.torproject.net
|
| |
|
|
| |
build-sunet-a.torproject.net
|
| |
|
|
|
| |
This is mainly for shipping a new version right after release as our alpha
release broke 0.0.13 (see: #23692).
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
By default NoScript attempts to find URLs in "javascript:.*" strings
found in <a> element href atributes and in <option> element value
attributes. When such links (or options) are clicked/selected,
NoScript attempts to navigate the page to said URL. These navigations
are treated as intternal requests, and get pacced onto the catch-all
circuit.
This behaviour can be turned off by disabling the 'noscript.fixLinks'
flag, so we do so for each Tor Browser build target's
extension-overrides.js.
|
| |
|
|
| |
README.HACKING
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
This reverts commit b6b18835b675dc1cae04faf474474d6a2ec65677.
|
| |
|
|
| |
https://docs.ansible.com/ansible/latest/porting_guide_2.0.html#deprecated
|
| |
|
|
| |
build-sunet-a.torproject.net
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
Versions bump and Changelog update
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
| |
We update the location to get the .dll as well taking Mike out of the
browser loop.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
| |
In the previous patch, we wrongly assumed that all runc 0.1.1 versions
exited with 0 when given an unknown command, and used that to detect the
version. It seems it is not true on all systems.
We are now using the 'runc --version' command to find the version of
runc.
We don't support versions older than 0.1.1.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The var_p/runc100 option is used to detect the version of runc we are
using. The format of the output of `runc --version` is different in both
versions, so it is not easy to parse. However I noticed that runc 0.1.1
does not exit with an error when we try to use a command that does not
exist while runc >= 1.0.0 does, so we use that to detect the runc
version we are using.
When using runc >= 1.0.0, we use `runc run` instead of `runc start':
https://github.com/opencontainers/runc/commit/c669b8d1568633c68bd915561ceb2e5ecc1bfc6a
In the runc config.json file, the format of the capabilities has been
changed. Instead of having one list of capabilities, there are now
separate lists for bounding, effective, inheritable, permitted and
ambient capabilities (and we use the same list for all of them):
https://github.com/opencontainers/runtime-spec/commit/eb114f057094dd2314682d55f8cb9c189915ac86
We also add /proc/timer_list and /sys/firmware to the "maskedPaths", as
those paths are now listed by default in the config.json file generated
by `runc spec` with runc 1.0.0.
|
| | |
|
| |
|
|
|
|
|
|
|
|
| |
In #23425 we redirected the output from runc start to /dev/null to
avoid annoying messages saying "stdin: is not a tty". However this also
removes stderr from build logs.
The command producing the `stdin: is not a tty` message is the `chown -R`
command from runc/remote_put, so we now only hide stderr from that
part while keeping it for the runc command running the build script.
|
| |
|
|
|
| |
With #23403 we started using the tor-browser-build/tmp directory for the
builds.
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Set process.terminal to false when building, but keep it to true when we
open a debugging shell.
https://github.com/opencontainers/runtime-spec/blob/master/config.md#process
To do that we update the runc config.json in runc/remote_exec rather
than only once in runc/remote_start.
We also redirect runc stderr to /dev/null to avoid some messages saying
"stdin: is not a tty" during the build.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
| |
In test builds only the en-US locale is provided, this 'ALL' in the
filename is confusing. And the nightlies are not being built multi-
lingual for a while now in tor-browser-bundle
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
build-sunet-a.torproject.net
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Shipping HTTPS-Everywhere based on the new WebExtensions model broke its
functionality on higher security levels as our JavaScript blocking does
not whitelist `moz-extension:` which is needed.
It turns out that vanilla Firefox 52 ESR is affected as well but differently:
Backorting https://bugzilla.mozilla.org/show_bug.cgi?id=1329731 does not help
us as NoScript is regulating JavaScript related things.
This fixes #21270 as well.
|
| |
|
|
| |
* FIXES #23166: https://bugs.torproject.org/23166
|
| | |
|
| | |
|