* Scan from line start when finding tag in tokenizer
This resolves a crash that can occur for invalid bytes like carriage
returns that are valid characters when not parsed from within literals.
There are potentially other edge cases this could resolve as well, as
the calling code for this function didn't account for any potential
'pending_invalid_tokens' that could be queued up by the tokenizer from
within another state.
* Fix carriage return crash in multiline string
Follow the guidance of #38:
> However CR directly before NL is interpreted as only a newline and not part of the multiline string. zig fmt will delete the CR.
Zig fmt already had code for deleting carriage returns, but would still
crash - now it no longer does so. Carriage returns encountered before
line-feeds are now appropriately removed on program compilation as well.
* Only accept carriage returns before line feeds
Previous commit was much less strict about this, this more closely
matches the desired spec of only allow CR characters in a CRLF pair, but
not otherwise.
* Fix CR being rejected when used as whitespace
Missed this comment from ziglang/zig-spec#83:
> CR used as whitespace, whether directly preceding NL or stray, is still unambiguously whitespace. It is accepted by the grammar and replaced by the canonical whitespace by zig fmt.
* Add tests for carriage return handling
One of the main points of for loops is that you can safety check the
length once, before entering the loop, and then safely assume that every
element inside the loop is in bounds.
In master branch, the safety checks are incorrectly intact even inside
for loops. This commit fixes it. It's especially nice with multi-object
loops because the number of elided checks is N * M where N is how many
iterations and M is how many objects.
This also makes another breaking change to for loops: in order to
capture a pointer of an element, one must take the address of array
values. This simplifies a lot of things, and makes more sense than how
it was before semantically.
It is still legal to use a for loop on an array value if the
corresponding element capture is byval instead of byref.
This strategy uses pointer arithmetic to iterate through the loop. This
has a problem, however, which is tuples. AstGen does not know whether a
given indexable is a tuple or can be iterated based on contiguous
memory. Tuples unlike other indexables cannot be represented as a
many-item pointer that is incremented as the loop counter.
So, after this commit, I will modify AstGen back closer to how @vexu had
it before, using a counter and array element access.
* Allow unbounded looping.
* Lower by incrementing raw pointers for each iterable rather than
incrementing a single index variable. This elides safety checks
without any analysis required thanks to the length assertion and
lowers to decent machine code even in debug builds.
- An "end" value is selected, prioritizing a counter if possible,
falling back to a runtime calculation of ptr+len on a slice input.
* Specialize on the pattern `0..`, avoiding an unnecessary subtraction
instruction being emitted.
* Add the `for_check_lens` ZIR instruction.
RunStep is supposed to auto-detect whether the intend is for
side-effects or for producing an output file. The auto-detection logic
was incorrect, and this commit fixes it.
I tested this manually locally. Automated testing will require a more
significant investment in the test harness, which I will work on in a
future enhancement.
closes#14666
Currently `zig cc`, when confronted with a linker argument it does
not understand, skips the flag and emits a warning.
This has been causing headaches for people that build third-party
software (including me). Zig seemingly builds and links the final
executable, only to segfault when running it.
If there are linker warnings when compiling software, the first thing we
have to do is add support for ones linker is complaining, and only then
go file issues. If zig "successfully" (i.e. status code = 0) compiles a
binary, there is instead a tendency to blaim "zig doing something
weird". (I am guilty of this.) In my experience, adding the unsupported
arguments has been quite easy; see #11679, #11875, #11874 for recent
examples.
With the current ones (+ prerequisites below) I was able to build all of
the CGo programs that I am encountering at $dayjob. CGo is a reasonable
example, because it is exercising the unusual linker args quite a bit.
Prerequisites: #11614 and #11863.
Tested by: created a "hello world" C file and compiled with:
zig cc -v main.c -Wl,--sort-section=name -o main
... and verified the `--sort-section=name` is passed to ld.lld.
Refs https://github.com/zigchroot/zig-chroot/issues/1
When the following is specified
```
$ zig build-obj -fcompiler-rt example.zig
```
the resulting relocatable object file will have the compiler-rt
unconditionally embedded inside.
```
$ nm example.o
...
0000000012345678 W __truncsfhf2
...
```
