This makes the following changes for i386:
long long and unsigned long long have 4 byte alignment on non-Windows
f64 (double) has 4-byte alignment on non-Windows
long double is 80 bits and has 4 byte alignment on mingw
long double on android is 64 bits, not 80: https://www.uclibc.org/docs/psABI-i386.pdfFixes#12453Fixes#12987
I think this may be helpful in the future when we might want to calculate this again at some other point.
It also makes it more clear that the other two functions below it are only required for this calculation.
* prep: output_buffer -> output_buffer_slice
* fix: truncate lines accurately
Currently, the code assumes a terminal width of 100.
If we look at what's printed for the last test:
```
Test [1/1] test "basic functionality"... [101/100] this is a really long name designed to activate the truncation code. let's fi...
```
No, it does not really work because the relevant part here is `"[101/100] this is a really long name designed to activate the truncation code. let's fi... "`,
which is 90 characters, but we expect 100 because that's the width that is assumed.
The reason is that it also measures **unprintable characters** (escape sequences) at least non-Windows systems.
With this commit the output is now:
```
Test [1/1] test "basic functionality"... [101/100] this is a really long name designed to activate the truncation code. let's find out if...
```
Of which `"[101/100] this is a really long name designed to activate the truncation code. let's find out if... "`
is the actual output of *our* `std.Progress` (remember that `zig test` has an `std.Progress` and our test itself does).
The length of that string is 100. Now the length is consistent with Windows where we don't use escape sequences. This issue was only present on non-Windows systems.
* feat: decide optimal maximum width
This is done by 1. getting the current terminal width and 2. subtracting that by the current cursor column. This accounts for previous output from someone else.
* test: add more tests
They make it easier to see how the progress line is printed in different cases.
* style: fix typo and improve docs
It also expands an acronym used as a variable name. It confused me.
* cleanup: import std.time
* test: add test
* fix: limit termios usage to Linux only for now
* fix: missing cast on Windows
* test: try to debug failure
* fix: fix off-by-one and disable tests
* docs: make comment clearer
* fix: more durability
* fix(getTerminalWidth): change order
Previously, you might obtain `-lLLVM-15` from the CMake configuration,
but we might not be able to locate the library if it's not in your
system library path.
build.zig:
- use "-I" instead of "-isystem" for `b.addSearchPrefix()`
main.zig:
- silently ignore superfluous search dirs
- warn when a dir is added to multiple searchlists
- consolidate "expected paramter after {s}" fatal error messages
- rename command-line switch `-dirafter` → `-idirafter`
closes#12888
Ideally this duplicated code could be factored out into a function, but
there doesn't seem to be any way in the Zig type system to represent an
argument to a function called at comptime that is only needed if it is
comptime-known. Instead, we document what is going on in an adjacent
comment in case it gets copy-pasted into new methods in the future.
When trying to allocate memory for functions like `Managed.init` and
`Managed.set` on the stack, a comptime-known allocation size is desired.
The doc comments for these functions indicate that `calcLimbLen` can be
used to determine how many limbs to allocate, but if `value` is not
comptime-known, then neither is `calcLimbLen(value)`. However, an upper
bound on the allocation size is still computable at comptime in this
case, so this note documents an expression that can be used, rather than
trying to add it to every doc comment that mentions `calcLimbLen`.
This allows converting a comptime_int to an optional integer type, which
either behaves the same as an implicit cast or produces null if the
argument is outside the range of the destination type.
This re-write was needed to fix deficiencies in the existing ldexp,
which was failing to compute correct results for both f16 and f80.
It would be nice to add a fast multiplication-based fallback in the
future for targets that have a hardware FPU, but this implementation
should be much faster than the existing for targets without one.
Windows requires the directory handle to be closed before attempting to delete the directory, so now we do that and then re-open it if we need to retry (from getting DirNotEmpty when trying to delete).
This was sized large so that `getdents` (and other platforms' equivalents) could provide large amounts of entries per syscall, but some benchmarking seems to indicate that the larger 8192 sizing doesn't actually lead to performance gains outside of edge cases like extremely large amounts of entries within a single directory (e.g. 25,000 files in one directory), and even then the gains are minimal ('./walk-8192 dir-with-tons-of-entries' ran 1.02 ± 0.34 times faster than './walk-1024 dir-with-tons-of-entries').
Note: Sizes 1024 and 2048 had similar performance characteristics, so the smaller of the two was chosen.
