Problem:
- Output of decode and encode grow as the decode and encode
happen. This is inefficient because a large size will require many
reallocations.
- `const` qualifiers are missing on variables which are not intended
to change.
Solution:
- Since the size of the decoded or encoded message is known prior to
starting, calculate the size and set the output to that size
immediately. All appends will not incur the reallocation overhead.
- Add `const` qualifiers to show intent.
Problem:
- Constructors and conversion operators are not `constexpr`,
but they can be.
- `constexpr` is needed here so that other classes can add `constexpr`
evaluation.
Solution:
- Add the `constexpr` keyword to the constructors and
conversion operators.
- Add `static_assert` tests which ensure the capability works.
String literals are just pointers to a constant character. It should be
possible to format them as such. (The default is to print them as
strings still.)
When we write the format specifier '{:#08x}' we are asking for eight
significant digits, zero padding and the prefix '0x'.
However, previously we got only six significant digits because the
prefix counted towards the width. (The number '8' here is the total
width and not the number of significant digits.)
Both fmtlib and printf shared this behaviour. However, I am introducing
a special case here because when we do zero padding we really only care
about the digits and not the width.
Notice that zero padding is a special case anyways, because zero padding
goes after the prefix as opposed to any other padding which goes before
it.
It's now save to pass a signed integer as parameter and then use it as
replacement field (previously, this would just cast it to size_t which
would be bad.)
I put this into the <AK/PrintfImplementation.h> header in the hope that
it could be re-used by the printf implementation. That would not be
super trivial though, so I am not doing that now.
With this commit, <AK/Format.h> has a more supportive role and isn't
used directly.
Essentially, there now is a public 'vformat' function ('v' for vector)
which takes already type erased parameters. The name is choosen to
indicate that this function behaves similar to C-style functions taking
a va_list equivalent.
The interface for frontend users are now 'String::formatted' and
'StringBuilder::appendff'.
Two things I hate about C++:
1. 'int', 'signed int' and 'unsigned int' are two distinct types while
'char, 'signed char' and 'unsigned char' are *three* distinct types.
This is because 'signed int' is an alias for 'int' but 'signed char'
can't be an alias for 'char' because on some weird systems 'char' is
unsigned.
One might think why not do it the other way around, make 'int' an
alias for 'signed int' and 'char' an alias for whatever that is on
the platform, or make 'char' signed on all platforms. But who am I
to ask?
2. 'unsigned long' and 'unsigned long long' are always different types,
even if both are 64 bit numbers.
This commit fixes a few bugs that coming from this.
See Also: 1b3169f405.
OutputMemoryStream was originally a proxy for DuplexMemoryStream that
did not expose any reading API.
Now I need to add another class that is like OutputMemoryStream but only
for static buffers. My first idea was to make OutputMemoryStream do that
too, but I think it's much better to have a distinct class for that.
I originally wanted to call that class FixedOutputMemoryStream but that
name is really cumbersome and it's a bit unintuitive because
InputMemoryStream is already reading from a fixed buffer.
So let's just use DuplexMemoryStream instead of OutputMemoryStream for
any dynamic stuff and create a new OutputMemoryStream for static
buffers.
Leverage constexpr and __builtin_ffs for Bitmap::find_first. Also add
a variant Bitmap::find_one_anywhere that can start scanning at a
provided hint.
Also, merge Bitmap::fill_range into the already existing Bitmap::set_range
Specifically:
- post-increment actually implemented pre-increment
- helper-templates that provided operator{+,-,*,/}() couldn't possibly work,
because the interface of add (etc) were incompatible (not taking a Checked<>,
and returning void)
In particular: consistent rounding and extreme values.
Before, rounding was something like 'away from 0.999...', which led to
surprising corner cases in which the value was rounded up.
Now, rounding is always 'down'.
This even works for 0xffffffff, and also for 0xffffffffffffffffULL on 64-bit.
Thankfully, this hasn't happened in any other code yet, but it happened
while I was trying something out. Using '==' on two ByteBuffers to check
whether they're equal seemed straight-forward, so I ran into the trap.
This seems to be because ByteBuffer implements 'operator bool', and C++
considers bool to be an integer type. Thus, when trying to find a way to
evaluate '==', it attempts integer promotion, which in turn finds 'operator bool'.
This explains why all non-empty buffers seem to be equal, but different from the
empty one. Also, why comparison seems to be implemented.
clang-format automatically sorts include statements that are in a
'block'. Adding a whitespace prevents this. It is crutial that
<AK/TestSuite.h> is included first because it redefines some macros.
