2015-02-05 18:58:48 +00:00
|
|
|
C++11 in Skia
|
|
|
|
=============
|
|
|
|
|
|
|
|
Skia is exploring the use of C++11. As a library, we are technically limited
|
|
|
|
by what our clients support and what our build bots support.
|
|
|
|
|
|
|
|
Skia may also be limited by restrictions we choose put on ourselves. This
|
|
|
|
document is not concerned with C++11 policy in Skia, only its technical
|
|
|
|
feasibility. This is about what we can use, a superset of what we may use.
|
|
|
|
|
|
|
|
The gist:
|
2015-02-08 15:09:22 +00:00
|
|
|
- C++11 the language as supported by GCC 4.4 or later is probably usable.
|
|
|
|
- If you break a bot, that feature is not usable.
|
|
|
|
- The C++11 standard library can't generally be used.
|
|
|
|
- Local statics are not thread safe.
|
2015-02-05 18:58:48 +00:00
|
|
|
|
|
|
|
|
|
|
|
Clients
|
|
|
|
-------
|
|
|
|
|
|
|
|
The clients we pay most attention to are Chrome, Android, Mozilla, and a few
|
|
|
|
internal Google projects.
|
|
|
|
|
|
|
|
Chrome builds with a recent Clang on Mac and Linux and with a recent MSVC on
|
|
|
|
Windows. These toolchains are new enough to not be the weak link to use any
|
|
|
|
C++11 language feature. But Chrome still supports Mac OS X 10.6, which does
|
|
|
|
not ship with a C++11 standard library. So [Chrome has banned the use of the
|
|
|
|
C++11 standard library](http://chromium-cpp.appspot.com/). Some header-only
|
|
|
|
features are probably technically fine, but the Mac toolchain will prevent us
|
|
|
|
from even trying at compile time as long as we target 10.6 as our minimum API
|
|
|
|
level.
|
|
|
|
|
|
|
|
Chrome intentionally disables thread-safe initialization of static variables,
|
|
|
|
and MSVC doesn't support it at all, so we cannot rely on that.
|
|
|
|
|
|
|
|
Android builds with either a recent GCC or a recent Clang. They're generally
|
|
|
|
not a weak link for C++11 language features. Android's C++ standard library
|
|
|
|
has always been a pain, but since we can't use it anyway (see Chrome), don't
|
|
|
|
worry about it.
|
|
|
|
|
|
|
|
Mozilla's current weak link is a minimum requirement of GCC 4.6. Most features
|
|
|
|
marked in red on Mozilla's C++11 [feature
|
|
|
|
matrix](https://developer.mozilla.org/en-US/docs/Using_CXX_in_Mozilla_code) are
|
|
|
|
marked that way because they arrived in GCC 4.7 or GCC 4.8. Their
|
|
|
|
minimum-supported Clang and MSVC toolchains are great. They also appear to ban
|
|
|
|
the C++ standard library.
|
|
|
|
|
|
|
|
Internal Google projects tend to support C++11 completely, including the
|
|
|
|
full C++11 standard library.
|
|
|
|
|
|
|
|
|
|
|
|
Bots
|
|
|
|
----
|
|
|
|
|
|
|
|
Most of our bots are pretty up-to-date: the Windows bots use MSVC 2013, the Mac
|
|
|
|
bots a recent Clang, and the Linux bots GCC 4.8 or a recent Clang. Our Android
|
|
|
|
bots use a recent toolchain from Android (see above), and our Chrome bots use
|
|
|
|
Chrome's toolchains (see above). I'm not exactly sure what our Chrome OS bots
|
|
|
|
are using, but they've never been a problem.
|
|
|
|
|
|
|
|
A few miscellaneous compile-only bots are actually our current overall weak link:
|
2015-02-08 23:36:19 +00:00
|
|
|
- Our NaCl builds use an old non-PNaCl toolchain, which is based on GCC
|
|
|
|
4.4. GCC 4.4 has some support for C++11, but it's not nearly complete.
|
|
|
|
There is no upgrade path except PNaCl; even the very latest NaCl toolchain
|
|
|
|
is GCC 4.4, while PNaCl is based on Clang 3.4 (with complete C++11 support).
|
|
|
|
- Our iOS builds are driven from a Mac 10.7 machine using some unknown old Clang.
|
|
|
|
Who knows how old that is or what it supports? It's probably due for an update.
|
2015-02-05 18:58:48 +00:00
|
|
|
|
|
|
|
If we were to eliminate the problems of the NaCl and iOS bots, our ability to
|
|
|
|
use C++11 would match Mozilla's list nearly identically.
|