[llvm-dev] LLVM Weekly - #275, April 8th 2019

[Alex Bradbury via llvm-dev]

LLVM Weekly - #275, April 8th 2019
==================================

If you prefer, you can read a HTML version of this email at
<http://llvmweekly.org/issue/275>.

Welcome to the two hundred and seventy-fifth issue of LLVM Weekly, a weekly
newsletter (published every Monday) covering developments in LLVM, Clang, and
related projects. LLVM Weekly is brought to you by [Alex
Bradbury](https://www.linkedin.com/in/alex-bradbury/). Subscribe to future
issues at <http://llvmweekly.org> and pass it on to anyone else you think may
be interested. Please send any tips or feedback to <asb at asbradbury.org>, or
@llvmweekly or @asbradbury on Twitter.

I'm at EuroLLVM today and tomorrow - be sure to say hi!


## News and articles from around the web

The April LLVM Austin social will [take
place](http://lists.llvm.org/pipermail/llvm-dev/2019-April/131632.html) this
Thursday 11th April at 6pm.

The Portable Computing Language (pocl) 1.3 has [been
released](http://lists.llvm.org/pipermail/llvm-dev/2019-April/131561.html),
this is an open source implementation of the OpenCL standard. This release
adds support for Clang/LLVM 8.0, amongst other changes.


## On the mailing lists

* Erich Focht published an RFC on [adding the NEC SX-Aurora VE backend to
upstream
LLVM](http://lists.llvm.org/pipermail/llvm-dev/2019-April/131580.html). It
plans to build on the LLVM-VP vector predication work.

* JF Bastien
[proposes](http://lists.llvm.org/pipermail/llvm-dev/2019-April/131486.html)
flipping the switch on mandating C++14 compliant compilers for building LLVM.
It sounds like most interested parties are ready, with Chandler Carruth
[indicating](http://lists.llvm.org/pipermail/llvm-dev/2019-April/131489.html)
Google needs a little more time.

* A case of poor performance when using `git log --parents` on the LLVM
monorepo is [likely to result in an upstream git
fix](http://lists.llvm.org/pipermail/llvm-dev/2019-April/131506.html).

* Puyan Lotfi shares an [RFC on
clang-ifso](http://lists.llvm.org/pipermail/cfe-dev/2019-April/061953.html), a
new clang-based tool for generating interface libraries.

* Bruno Ricci starts a discussion on [improving Clang serialization
tests](http://lists.llvm.org/pipermail/cfe-dev/2019-April/061896.html).

* Don Hinton's brief post on [adding
`isa_or_null<>`](http://lists.llvm.org/pipermail/llvm-dev/2019-April/131555.html)
generated a relatively large number of responses.

* Peter Hosek
[proposes](http://lists.llvm.org/pipermail/llvm-dev/2019-April/131594.html)
some cleanups to the compiler-rt builtins, reformatting and adjusting comment
style to match the LLVM coding standards.


## LLVM commits

* A new `isa_and_nonnull<>` operator was added which works just like `isa<>`,
but allows for a null pointer as an argument.
[r357761](https://reviews.llvm.org/rL357761).

* New PDB documentation was added.
[r357777](https://reviews.llvm.org/rL357777),
[r357788](https://reviews.llvm.org/rL357788),
[r357826](https://reviews.llvm.org/rL357826).

* The RISC-V LLVM backend learned to emit appropriate address sequences for
`-mcmodel=medium`. [r357393](https://reviews.llvm.org/rL357393).

* A new `optForNone()` method was added to Function to more easily test for
`-O0`. [r357638](https://reviews.llvm.org/rL357638).

* SelectionDAG::computeKnownBits can now compute known bits for
`ISD::CopyFromReg`. [r357745](https://reviews.llvm.org/rL357745).

* The X86 CMOV, SETcc, and Jcc instruction definitions were refactored to merge
the condition codes into single instructions storing the condition code as an
immediate. [r357800](https://reviews.llvm.org/rL357800),
[r357801](https://reviews.llvm.org/rL357801),
[r357802](https://reviews.llvm.org/rL357802)


## Clang commits

* Transformer, a library for clang-based source-to-source transformations was
committed. [r357768](https://reviews.llvm.org/rL357768).

* For loops with more than one transformation, the order in which those
transformations are applied will now be encoded in the emitted metadata.
[r357415](https://reviews.llvm.org/rL357415).


## Other project commits

* libc++'s handling of error flags and exceptions propagated from input stream
operations was modified to match behaviour of other implementations.
[r357775](https://reviews.llvm.org/rL357775).

* LLDB has started to move over to using the minidump parser in
llvm/Object rather than its own. [r357748](https://reviews.llvm.org/rL357748).