We were caching the ConstExprValue of string literals,
which works if you can never modify ConstExprValues.
This premise is broken with `comptime var ...`.
So I implemented an optimization in ConstExprValue
arrays, where it stores a `Buf *` directly rather
than an array of ConstExprValues for the elements,
and then similar to array of undefined, it is
expanded into the canonical form when necessary.
However many operations can happen directly on the
`Buf *`, which is faster.
Furthermore, before a ConstExprValue array is expanded
into canonical form, it removes itself from the string
literal cache. This fixes the issue, because before an
array element is modified it would have to be expanded.
closes#1076
from #llvm IRC:
<andrewrk> does llvm 7 as a .so have some kind of new initialization
requirement? I'm getting a segfault in llvm::DIBuilder::createFile
(with valid non-null parameters), when linking my frontend against
llvm-7.so but not when linking against llvm .a libraries
<d0k> we have an ABI bug in LLVM 7 when the .so is built with gcc
but your program is build with clang. I'm sorry for that.
this prevents the situation where we determine the cache
manifest and write it, but then crash or otherwise error out
before putting the artifacts in the proper place.
now the artifacts will be in place because cache_release
happens after that step is done.
init-lib creates a working static library with tests, and
init-exe creates a working hello world with a `run` target.
both now have test coverage with the new "cli tests" file.
closes#1035
To make dump testable added dumpToSteam which takes a stream as input
and added the stream as a paraemter to dumpRecursive.
Added test "std.atomic.Queue dump"
And to make the test more robust SliceOutStream.pos is now public. This
allows the user of SliceOutStream to know the length of the data captured.
