Field width conversion `.*` always takes an `int`.
On Win32, casting `long` to `int` triggers a `-Wuseless-cast` warning.
Omitting the cast works correctly.
On Linux/amd64, casting `long` to `int` works correctly.
Omitting the cast triggers a `-Wformat` warning.
Add a macro that conditionally expands to `static_cast<int>` or to ``,
as necessary for the target platform.
C casts do not require parentheses. C++ casts require grouping around
the target. Prepare for conversion to C++ casts by adding otherwise
unnecessary parentheses around the target of simple C casts.
s/(\s*unsigned\s\+char\s*\(\*\+\)\s*)\s*\([&+-]\?\)\([[:alnum:]_.]\+\s*->\s*\)*\([[:alnum:]_.]\+\)\(\s*\([];+>)*\/^%,|&<>]\)\|$\|\(\s*-\s*[^>]\)\)/(uint8_t \1)(\2\3\4)\5/g
Uses of `(short*)` usually want exactly a 16-bit signed integer. Most
platforms provide that as `short`, but that is not guaranteed.
s/(short \(*\+\)\s*)/(int16_t \1)/g
This pass only targets commonly used standard types.
s/(\(\s*\(\(un\)\?signed\|int\|char\|short\|long\|float\|double\|s\?size_t\|\(u\?int[[:digit:]]\+_t\)\)\)\s*)\s*(/static_cast<\1>(/g
C casts do not require parentheses. C++ casts require grouping around
the target. Prepare for conversion to C++ casts by adding otherwise
unnecessary parentheses around the target of some C casts.
This pass attempts to process expressions that involve parenthesized or
bracketed subexpressions, but only if those subexpressions do not
themselves contain parenthesized or bracketed subexpressions.
(int) f(1); // changed
(int) f(g()); // not changed
perl -p -i -e 's/(\(\s*((?:un)?signed|int|char|short|long|float|double|s?size_t|(?:u?int[[:digit:]]+_t))\s*\**\s*\)\s*)([&+-]?)([[:alnum:]_.]+\s*->\s*)*([[:alnum:]_.]+)((?:\s*(?:\[[^][]*\])*|(?:\([^()]*\))*))(\s*([;+>*\/^%,|&<>])|$|(\s*-\s*[^>]))/\1\(\3\4\5\6\)\7/g'
This pass only targets commonly used standard types.
s/(\(\s*\(\(un\)\?signed\|int\|char\|short\|long\|float\|double\|s\?size_t\|\(u\?int[[:digit:]]\+_t\)\)\)\s*)\s*(/static_cast<\1>(/g
C casts do not require parentheses. C++ casts require grouping around
the target. Prepare for conversion to C++ casts by adding otherwise
unnecessary parentheses around the target of simple C casts.
This pass does not attempt to process expressions that involve
any subexpression that can nest arbitrarily, such as parentheses or
brackets. It also works only on commonly used standard types.
(int) a->b; // changed
(int) a[b]; // not changed
s/\((\s*\(\(un\)\?signed\|int\|char\|short\|long\|float\|double\|s\?size_t\|\(u\?int[[:digit:]]\+_t\)\)\s*\**\s*)\s*\)\([&+-]\?\)\([[:alnum:]_.]\+\s*->\s*\)*\([[:alnum:]_.]\+\)\(\s*\([];+>)*\/^%,|&<>]\)\|$\|\(\s*-\s*[^>]\)\)/\1(\5\6\7)\8/g
Mac OS X defines uint_fast32_t to unsigned int, causing ambiguous
overloads between RAIIdmem::operator[](std::size_t) and
RAIIdmem::operator[](int). Adding a disambiguating overload for OS X
breaks Windows. Remove operator pointer and operator[]. Rely on the
inherited operator[] for indexing. Require users to call ->get() to
convert to a simple pointer.
First btb ambiguity reported: https://github.com/dxx-rebirth/dxx-rebirth/pull/34
Second btb ambiguity reported: https://github.com/dxx-rebirth/dxx-rebirth/pull/43
C++ does not require this pattern.
import re, fileinput
to = re.compile(r'^typedef struct ([a-z_A-Z]+)\s*{')
tc = re.compile(r'^}(.*?)\s*([a-z_A-Z]+);$')
osn = None
for line in fileinput.input(inplace=True):
m = to.match(line)
if m:
osn = m.group(1)
print 'struct %s\n{' % osn
continue
if osn:
m = tc.match(line)
if m:
csn = m.group(2)
if osn == csn:
print '}%s;' % m.group(1)
osn = None
continue
else:
osn = None
print line,
