llvm dev - Oct 2013 - [LLVMdev] [cfe-dev] RFC: A proposal to move toward using C++11 features in LLVM & Clang / bounding support for old host compilers

On 10/29/13 06:52 AM, Chandler Carruth wrote:
>
> On Mon, Oct 28, 2013 at 4:47 PM, "C. Bergström" 
> <cbergstrom at pathscale.com <mailto:cbergstrom at
pathscale.com>> wrote:
>
>     For those driving c++11 in clang/llvm - Would it generally be
>     acceptable to have a "sunrise" period where the preliminary
>     evaluation has been done (buildbots, compiler evaluate.. etc) and
>     the 1st actual c++11 commit hits the repo. (30-60 days?)
>
>
> I really don't think we need this level of complexity to the planning. 
> I think we can give a heads up at a high level (as already discussed), 
> check the build bots are updated, and flip the switch.
fuzzy://How much "heads up"
>
>     -------------
>     My concern/thoughts - When we swap out STDCXX for libc++ - We
>     aren't able to self host clang.
>
>
> No part of this relies on using libc++ on Linux instead of libstdc++. 
> That's an orthogonal issue which I'm not even attempting to
address.
>
> Certainly, the platforms with only libc++ are self hosting clang 
> successfully today.
Hmm....
>
>     This could be entirely *our* fault, but it hasn't been
>     investigated extensively. (We also see Perennial C++ testsuite
>     regressions which appear to come from libc++, but also not
>     investiaged/confirmed) Having a sunrise period would allow us to
>     investigate this as well as report any potentially blocking problems.
>
>     Having a gnu-free self hosting[1] policy attached to this would
>     also be great - that makes a potentially easier backup solution to
>     anyone on [linux] with older gnu compilers
>
>
> Absolutely not. I really don't want to tie this to more things. If you 
> think we should pursue such a goal, fine, but it is independent of the 
> decision about using C++11, and not something I'm currently striving
for.
-1
If we do it - we should do it properly.
---------
I'm strongly against anything which forces this to rely on gnu runtime 
and libs.
---------
Further - While c++11 is the best thing ever - Is there actually a 
driving use case where using c++11 features will significantly improve 
the quality of the codebase? implementation perspective or performance? 
Are there coding style guidelines which should be updated - I've seen 
code (ab)using auto where it was mostly author preference and not really 
necessary. You really push for this, but the underlying motivation still 
seems unclear. (Show me the c++11 code in the subproject which is the 
root cause please)

On Mon, Oct 28, 2013 at 5:06 PM, "C. Bergström" <cbergstrom at
pathscale.com>wrote:
> fuzzy://How much "heads up"

One full release cycle, so approximately 6 months before a release
incorporating this. In addition, probably several weeks to a month between
the initial branch for the 3.4 release and the commits that make this so.
Still more weeks between when we put it into the release notes for 3.4 and
blog about it.

> -1
> If we do it - we should do it properly.
>
We're not talking about doing it improperly, but about doing two
independent things independently.

> ---------
> I'm strongly against anything which forces this to rely on gnu runtime
and
> libs.
>
You are not being forced to rely on anything. Many people have successfully
bootstrapped with libc++, and if you find issues with your setup doing such
a bootstrap, patches to fix that would be very welcome.

And moreover, a requirement to not use GNU runtimes or libraries in the
bootstrap would be totally novel for the project. LLVM and Clang require
xlC and MSVC to bootstrap on certain platforms, and you're OK with that?
There are plenty of platforms with no Clang port at all but where GCC works
fine, and we rely on a GCC host there. This is not new, and not something
we should burden the discussion about C++11 features with.


> ---------
> Further - While c++11 is the best thing ever - Is there actually a driving
> use case where using c++11 features will significantly improve the quality
> of the codebase?

Yes. See my first email for some of them. See numerous other discussions.
Have you tried writing any code with C++11?

> implementation perspective or performance?

We get most of this already today, but it would be cleaner if we just
admitted we rely on it.

> Are there coding style guidelines which should be updated - I've seen
code
> (ab)using auto where it was mostly author preference and not really
> necessary.

Yes, and we have code review for this. If you would like to help review
patches for high quality usage of C++11 constructs, that would be welcome.

> You really push for this, but the underlying motivation still seems
> unclear. (Show me the c++11 code in the subproject which is the root cause
> please)
>
I don't know how it could be unclear. I spelled out reasons in my first
email. There was a more thorough discussion in past threads on this
subject. I'm not going to repeat myself here.

And it should be quite obvious that LLD is actively using C++11 code to
good effect? That was discussed extensively when the project was added.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20131028/8e2b8988/attachment.html>

On 10/29/13 07:27 AM, Chandler Carruth wrote:
> On Mon, Oct 28, 2013 at 5:06 PM, "C. Bergström" 
> <cbergstrom at pathscale.com <mailto:cbergstrom at
pathscale.com>> wrote:
>
>     fuzzy://How much "heads up"
>
>
> One full release cycle, so approximately 6 months before a release
If it's 3-6 months from *today* before something hits clang svn trunk 
that should be enough time to address any problems. (Such as testing 
libc++ more extensively and or sending any patches if necessary) I don't 
know the 3.4 release timeline and if you want to adopt this immediately 
after that has branched. [My other points below can be ignored if this 
is true]






---------------------------------
> incorporating this. In addition, probably several weeks to a month 
> between the initial branch for the 3.4 release and the commits that 
> make this so. Still more weeks between when we put it into the release 
> notes for 3.4 and blog about it.
>
>     -1
>     If we do it - we should do it properly.
>
>
> We're not talking about doing it improperly, but about doing two 
> independent things independently.
>
>     ---------
>     I'm strongly against anything which forces this to rely on gnu
>     runtime and libs.
>
>
> You are not being forced to rely on anything. Many people have 
> successfully bootstrapped with libc++, and if you find issues with 
> your setup doing such a bootstrap, patches to fix that would be very 
> welcome.
>
> And moreover, a requirement to not use GNU runtimes or libraries in 
> the bootstrap would be totally novel for the project. LLVM and Clang 
> require xlC and MSVC to bootstrap on certain platforms, and you're OK 
> with that? There are plenty of platforms with no Clang port at all but 
> where GCC works fine, and we rely on a GCC host there. This is not 
> new, and not something we should burden the discussion about C++11 
> features with.
On linux amd64/intel64 is what I meant by "regression". The other 
platforms you reference are still work-in-progress. It wasn't my 
intention to include them as a blocker - (just to
clarify)
>
>     ---------
>     Further - While c++11 is the best thing ever - Is there actually a
>     driving use case where using c++11 features will significantly
>     improve the quality of the codebase?
>
>
> Yes. See my first email for some of them. See numerous other 
> discussions. Have you tried writing any code with C++11?
Cherry picking the points from your original email.
>
> Some notable features we would get to use:
>
> - r-value references, move semantics, etc
> - auto
> - range for loops
> - lambdas
> - static_assert
> - nullptr
> - std::unique_ptr, std::tuple, some other nice library stuff
>
>
I'd love to see/review an updated style guideline before anything hits 
SVN trunk so that it's super clear on what's allowed. Maybe I'm too 
conservative, but it's still fuzzy on how using some of those features 
would benefit the compiler internally.

Usage of c++11 code in applications (chrome) and the compiler are 2 
different worlds to me. You ask if I write c++11 codes - No, but 
increasingly I do see it being used in production and forced to deal 
with it. Sometimes it makes sense and sometimes I think meh or wtf -

If it doesn't make the code:
     a. easier to read
     b. easier to write
     c. higher performance

it has no value or negative value.

--------------------------
I don't mean to nit-pick||troll and I'm not opposed to this idea at all 
- I just like to have specifics and facts. (Please feel free to ignore 
these questions below)
> It also would have the benefit of removing divergence between LLVM 
> sub-projects already using C++11 features
Which sub-projects would this benefit?
>
> There is increasing pressure for LLVM to make use of new C++ language 
> and library features.
pressure from where? (sub-projects, google.. ?)
>
> I have historically been more conservative on this topic, but 
> listening to many people and looking at some of the C++ features we 
> are missing
watercooler or mailing list?
>
>
> These days, this list seems increasingly worth the cost of forcing 
> users to get a modern toolchain onto their systems.
>
I'm all for modern toolchains, but a modern toolchain requirement and 
depending on c++11 features are independent issues. You could make a 
policy that ______ doesn't care about building and supporting gcc-3.x on 
________ old ${platform} - I'm cool with that, but it doesn't mean we 
must rely on c++11 in the process.

I agree that a reasonable amount of notice should be given before making the
switch on top-of-trunk.
Quite a few out-of-tree projects try to stay close to top-of-trunk, and I think
that’s a good thing, since it provides quick detection & feedback on
regressions breaking out-of-tree projects.


To increase the likelihood that out-of-tree projects keep on following
top-of-trunk closely, I think it’s necessary to also clearly announce (on this
list) well in advance (e.g. at least one month, ideally more) when the minimum
requirements on tools needed to build LLVM change.

 

Thanks,

 

Kristof

 

From: llvmdev-bounces at cs.uiuc.edu [mailto:llvmdev-bounces at cs.uiuc.edu] On
Behalf Of Chandler Carruth
Sent: 29 October 2013 00:27
To: C. Bergström
Cc: clang-dev Developers; LLVM Developers Mailing List
Subject: Re: [LLVMdev] [cfe-dev] RFC: A proposal to move toward using C++11
features in LLVM & Clang / bounding support for old host compilers

 

On Mon, Oct 28, 2013 at 5:06 PM, "C. Bergström" <cbergstrom at
pathscale.com> wrote:

fuzzy://How much "heads up"

 

One full release cycle, so approximately 6 months before a release incorporating
this. In addition, probably several weeks to a month between the initial branch
for the 3.4 release and the commits that make this so. Still more weeks between
when we put it into the release notes for 3.4 and blog about it.

 

-1
If we do it - we should do it properly.

 

We're not talking about doing it improperly, but about doing two independent
things independently.

 

---------
I'm strongly against anything which forces this to rely on gnu runtime and
libs.

 

You are not being forced to rely on anything. Many people have successfully
bootstrapped with libc++, and if you find issues with your setup doing such a
bootstrap, patches to fix that would be very welcome.

 

And moreover, a requirement to not use GNU runtimes or libraries in the
bootstrap would be totally novel for the project. LLVM and Clang require xlC and
MSVC to bootstrap on certain platforms, and you're OK with that? There are
plenty of platforms with no Clang port at all but where GCC works fine, and we
rely on a GCC host there. This is not new, and not something we should burden
the discussion about C++11 features with.

 

 

---------
Further - While c++11 is the best thing ever - Is there actually a driving use
case where using c++11 features will significantly improve the quality of the
codebase?

 

Yes. See my first email for some of them. See numerous other discussions. Have
you tried writing any code with C++11?

 

implementation perspective or performance?

 

We get most of this already today, but it would be cleaner if we just admitted
we rely on it.

 

Are there coding style guidelines which should be updated - I've seen code
(ab)using auto where it was mostly author preference and not really necessary.

 

Yes, and we have code review for this. If you would like to help review patches
for high quality usage of C++11 constructs, that would be welcome.

 

You really push for this, but the underlying motivation still seems unclear.
(Show me the c++11 code in the subproject which is the root cause please)

 

I don't know how it could be unclear. I spelled out reasons in my first
email. There was a more thorough discussion in past threads on this subject.
I'm not going to repeat myself here.

 

And it should be quite obvious that LLD is actively using C++11 code to good
effect? That was discussed extensively when the project was added.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20131029/ecb42423/attachment.html>