Python 2 permitted defining __new__, then deleting it once it had been
used. Python 3 does not directly reject that pattern, but instead
always fails with an error about wrong number of arguments to object().
Switch to a simple classmethod that is explicitly invoked by the first
__init__().
In Python 3, filter returns an object that can be iterated to get the
filtered content. However, there is no way to reset the object, so it
is useless for persistence. Spend the one iteration copying the data
into a tuple, which can be iterated multiple times.
Python 3 `dict.keys()` returns an object that cannot be added to other
instances of itself. Pass the returned objects as a tuple, then let the
callee use itertools.chain to combine them during iteration.
Python 3 removed `dict.keys` and `dict.items`, then renamed
`dict.iterkeys` and `dict.iteritems` to `dict.keys` and `dict.items`,
respectively. The iteration semantics are preferable here, so add shims
to use the iteration API under a version-appropriate name.
Change PreservedEnvironment.__getitem__ back to a class method, instead
of an instance variable. Python 3 refuses to search the instance
dictionary for __getitem__, so the optimization of copying
`self.flags.__getitem__` to `self.__getitem__` breaks in Python 3.
Python 3 changed the rules for list comprehensions, so the comprehension
can no longer see a class-scope variable without a class qualifier. The
class qualifier is not legal until the class is finished. Work around
this limitation by expanding the use inline.
OpenBSD 6 uses `openbsd6` as its sys.platform. Strip trailing numbers,
then remove the hack for converting `linux2` into `linux`. This should
prevent compatibility problems when OpenBSD 7 is used.
Commit 265af2b9e5 added an extra test for whether the header could be
preprocessed, but reused the string shown when an empty test program is
compiled. This can confuse users since the same message is shown twice.
Differentiate the messages.
Fixes: 265af2b9e5 ("Extend SConf diagnostics for failed system headers")
gcc-4.9 warns (and due to -Werror, errors) for `png_time pt{}` because
the members are not explicitly initialized. This code is compiled, but
never run, so the explicit initialization is unnecessary. Later
versions of gcc accept both versions of this code without complaint.
Fixes: 131c1b9f4d ("Add support for PNG screenshots")
This is a longstanding cosmetic issue. Appending a tuple to a list
retains the tuple structure, which is then printed later. Append it as
a list so that it is flattened.
OpenBSD clang does not handle `-print-prog-name=as` in a useful manner.
Since this query is purely diagnostic, add an undocumented knob to
suppress the query.
thfrwn reports that OpenBSD (almost) works as-is, except that its
Python `sys.platform` is not in the allowed values list. Add it to the
list. By happy accident, unhandled `host_platform` is treated as
`'linux'` and `'linux'` produces the right result for OpenBSD.
Requested-by: thfrwn <https://github.com/dxx-rebirth/dxx-rebirth/pull/359>
Internally, hats are treated as a group of 4 buttons. Disabling all
buttons while not disabling all hats caused a build error. Since no one
reported this, this configuration is likely not used. Make the simple
fix of coercing max_hats_per_joystick to 0 when max_axes_per_joystick is
0.
psyke83 reports that Debian distributions for Raspberry Pi now require a
different name for some video libraries. Merge psyke83's SCons
enhancements to add support for the new names while retaining support
for the old names.
ignorecase=1 coerces the user's input only for validation, but retains
the original case for later processing.
ignorecase=2 coerces the user's input for all purposes, and retains the
coerced form for later processing.
Since other code compares the `raspberrypi` variable to specific values
from the whitelist, user input must be mapped to that whitelist both for
SCons validation and for that later code. With ignorecase=1, a user
could set `raspberrypi=MESA`, which would pass SCons validation, but
then not be treated as equal to `mesa` when computing default values.
With ignorecase=2, `raspberrypi=MESA` will be recorded as `mesa` and
compute the desired default value.
Fixes: 166b1ecd4d ("RPI: update vendor library names & add Mesa VC4 build support")
PhysFS 2.0 only offers PHYSFS_read/PHYSFS_write for I/O. PhysFS 3.0
deprecates PHYSFS_read / PHYSFS_write and offers PHYSFS_readBytes /
PHYSFS_writeBytes. Converting Rebirth to use the new API is somewhat
invasive, and would require dropping support for PhysFS 2.0.
For now, for compatibility with PhysFS 2.0, disable the deprecation
errors and continue to use the older functions.
Reported-by: gabeotisbenson <https://github.com/dxx-rebirth/dxx-rebirth/issues/352>
Vendor library has new names to avoid conflicts with Mesa. Update
raspberrypi argument to use new libraries and support Mesa VC4 driver.
* raspberrypi=1 will build against the (newly named) vendor libraries
* raspberrypi=mesa will target building against the VC4 driver.
Both will select GLES by default, but you can target GL for the VC4 driver
via "raspberrypi=mesa opengles=0"
This code was part of a feature abandoned before retail. It cannot be
usefully used in campaigns. Remove it to reduce code size and simplify
later changes.
SCons tries to probe the compiler designated by CXX before Environment()
returns. Set CXX in the call to Environment so that it probes the
correct tool.
If resolving the linker path fails, ld_path is not usable to run an
external program, but DXXProgram tried to use it as such. Test ld_path
before using it.
Splitting auto-generated paths is wasteful, but almost always harmless
on Linux. However, Windows uses backslash as a path separator, which
conflicts with its standard meaning as an escape character. Using
shlex.split on a generated Windows path strips required backslashes,
causing a later failure to find the command. Move the shlex.split call
to apply only to user-specified paths.
Reported-by: ef314159 <https://github.com/dxx-rebirth/dxx-rebirth/issues/349>
ef314159 reports that running SCons on Windows fails with an
AttributeError exception. `os.uname` is not available on Windows, but
SConstruct assumed it was. Trap the error and report a machine of
`None`, since this is purely for logging.
Reported-by: ef314159 <https://github.com/dxx-rebirth/dxx-rebirth/issues/348#issuecomment-334657171>
Fixes: 1ed7cec714 ("Tighten __builtin_constant_p check to handle gcc-7")
When print_function is imported (or Python 3 is used), print without
parentheses is an error. Historically, SCons has used only Python 2 and
has not enabled print_function. This may change, so add parentheses to
the one affected site. None of the print_function features are used, so
this should work in both Python 2 and Python 3.
LTO allows the compiler to prove variable values that non-LTO mode
cannot prove. The workaround for the gcc-7 __builtin_constant_p
quirk incorrectly caused LTO on older gcc to report __builtin_constant_p
as unsupported. Fix this by adding a stub asm() statement to hide the
value of the input parameter from the compiler.
Fixes: 1ed7cec714 ("Tighten __builtin_constant_p check to handle gcc-7")
Early implementations of integer_sequence used a naive implementation
that required one level of template depth per additional integer in the
sequence. Rebirth uses a private alternate implementation named
make_tree_index_sequence that requires only log(N) steps for an
N-element index_sequence. Recent versions of gcc ship a log(N) version
of integer_sequence. Probe for that version and, if found, use it
instead of the private implementation, on the theory that the compiler
writers did at least as good a job as I did, and possibly better if they
were able to leverage compiler implementation details.
- Add wrappers for PHYSFS_read and its convenience functions. Poison
the memory before calling PHYSFS, so that any uninitialized bytes
(likely caused by a short read) are reported if the caller tries to
access the value anyway.
- Add SConstruct options to enable wrapping, so that users do not need
to enumerate the wrapped functions manually.
- Fix link failure when using LTO+wrappers.
It is not meant to be included in every file. Starting in gcc-7, static
functions declared in valptridx.tcc generate -Wunused-function warnings
if the including file does not instantiate any templates that use the
static function. Avoid these warnings by including the tcc only in a
file that is guaranteed to instantiate templates that use these
functions.
gcc-7 warning -Wformat-truncation would be good to fix, but the compiler
does not offer an acceptable solution yet.
gcc-7 warning -Wimplicit-fallthrough requires more analysis of the
diagnosed sites to identify which, if any, require a resolution other
than suppressing the warning.
gcc-6 issues a "used but never defined" warning for functions which are
declared, never defined, and used only in contexts which are later
deleted by the optimizer. If the function were actually used, the link
would fail. Switch to a complicated mechanism to define the function
while keeping it out of line so that __attribute__((__error__)) is
triggered as intended.
Compile time sanity checks are built on gcc's __builtin_constant_p.
This intrinsic is useful for optimization and for some simple forms of
compile-time checks (such as the infamous open(2) missing mode), but
upstream does not guarantee that it will work reliably for more
complicated checks. Starting in gcc-7.1, __builtin_constant_p returns
an incorrect result of true for the vm_vec_sub expression
`(a.x == b.x && a.y == b.y && a.z == b.z)`
even when the blamed sites clearly cannot prove that the inputs are
equal. The useful result would be to return true if, and only if, the
inputs were provably identical; inputs which might be identical at
runtime, or might not, would return false. Based on a bug filed with
gcc and the developer comments there, it appears many projects have
assumed this intrinsic is usable in this way, but the gcc developers do
not guarantee that it can be used this way. Additionally, they believe
affected projects are rare and were wrong to use this intrinsic, so they
have no plans to fix this regression.
For more details, see gcc bug #72785
<https://gcc.gnu.org/bugzilla/show_bug.cgi?id=72785>. For some more
pointed commentary on this change in gcc, see Linus' kernel commit
474c90156c8dcc2fa815e6716cc9394d7930cb9c
<https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=474c90156c8dcc2fa815e6716cc9394d7930cb9c>.
Update the SConf test to include the reproducer shown by Markus
Trippelsdorf in gcc bug #72785, comment #0. This reproducer compiles
cleanly on <gcc-7, causing no change in semantics on older compilers.
Affected versions of gcc-7 will miscompile this reproducer into a link
error, causing SConf to record that the compiler does not optimize
__builtin_constant_p. This is pessimistic, since even affected versions
of gcc-7 can handle some simple uses of __builtin_constant_p correctly.
This is a quick fix to get gcc-7 users working. Upstream seems
disinclined to revert to the more useful semantics of <gcc-7 or to
introduce an alternative intrinsic with more helpful semantics. As a
minor enhancement for Rebirth, it would be nice to probe the limits of
gcc-7's handling of __builtin_constant_p so that cases it handles
correctly could be enabled for gcc-7 users, while still blacklisting the
more complicated checks that gcc-7 miscompiles.
Thanks to Markus Trippelsdorf for providing a minimal reproducer to
detect the affected gcc versions.
Reported-by: parkerlreed <https://github.com/dxx-rebirth/dxx-rebirth/issues/337>
When Rebirth is configured to use OpenGL, it cannot start with an SDL
that lacks OpenGL support. Detect this mismatch at build time. Require
the user to resolve the conflict by disabling Rebirth OpenGL or by
enabling SDL OpenGL.
The Windows implementation of inet_ntop incorrectly omits the `const`
qualifier on the input address. This broke the Windows build, since
Rebirth passes a const-qualified input address, as permitted by POSIX.
Some Windows cross-compilers, such as mingw32, lack a definition of
inet_ntop entirely. For such environments, fall back to inet_ntoa and
disallow building with IPv6.
Rework the formatting and add a Windows-specific test for whether
inet_ntop is available. As inet_ntop is specified by POSIX, assume all
modern non-Windows platforms support it.
SConstruct prefers to refer to linker flags for the main executable as
LINKFLAGS, not LDFLAGS. Rename the internal storage to eliminate a
special case in accessing it. Continue to use the environment variable
$LDFLAGS as an initial value, since many tools expect to pass linker
flags through $LDFLAGS.
zicodxx reported a compiler error when using inheriting constructors
when using an unspecified environment. Investigation shows that gcc-6.x
has a regression that causes spurious -Wuseless-cast warnings where the
compiler automatically synthesizes a cast, then warns the user that the
synthesized cast is useless.
Compensate for this by extending the -Wuseless-cast testcase to provoke
this bug, if present. Affected users will lose -Wuseless-cast, but
their build will succeed.
Reported-by: zicodxx <https://github.com/dxx-rebirth/dxx-rebirth/issues/316>
Popular Windows tools for stack unwinding lack support for the DWARF
debug format. Unwinding the stack without DWARF is unreliable without
frame pointers. Default frame pointers to enabled for Windows so that
these tools work despite their lack of DWARF support.
These flags are passed through to the underlying build, but not checked
by the SConf tests. They are meant for options which will work in the
main build, but fail in the configure environment, such as `-Wl,--wrap`.
Move DXX_VERSION_SEQ from DXXProgram to DXXCommon to make it visible to
DXXArchive.configure_environment(). If pch is enabled, define
DXX_VERSION_SEQ in dxxsconf.h and remove it from the command line. This
fixes a build failure where gcc ignored ownpch.cpp.gch because it was
built without DXX_VERSION_SEQ, but some files were built with
DXX_VERSION_SEQ.
To recognize getaddrinfo as supported, require that the AI_* flags
passed to it be usable:
- AI_NUMERICSERV must be defined as a number or be undefined
- If ipv6=1, then AI_V4MAPPED and AI_ALL must be defined as numbers
When getaddrinfo is used, raise _WIN32_WINNT from 0x501 to 0x600.
Otherwise, lower it to 0x500.
derhass reports that SDL on Raspbian provokes a warning from
-Wuseless-cast. Add a test that will report -Wuseless-cast as unusable
when SDL causes this warning. Developers on Raspbian will not get the
benefit of -Wuseless-cast, but their build will succeed.
Reported-by: derhass <https://github.com/dxx-rebirth/dxx-rebirth/issues/257>
Extended identifiers are convenient when supported, but derhass reports
that the latest stable gcc on Raspbian stable fails the test for
mandatory extended identifier support. Replace the Cxx11RequiredFeature
test for extended identifiers with an SConf test to selectively enable
-fextended-identifiers.
Reported-by: derhass <https://github.com/dxx-rebirth/dxx-rebirth/issues/254>
Fixes: a8a41aa2ff ("Move git version into Python class")
Previously, raspberrypi mode forced DXX_WORDS_NEED_ALIGNMENT on and any
other mode forced it off. Add a specific user setting for
DXX_WORDS_NEED_ALIGNMENT. Set its default as bool(raspberrypi) to
preserve the previous special case.
Rename symbol WORDS_NEED_ALIGNMENT to DXX_WORDS_NEED_ALIGNMENT to show that it is a DXX
symbol, not one inherited from a library.
This is a mostly automated transform, but the changes to SConstruct were
manual.
git grep -wl WORDS_NEED_ALIGNMENT -- '*.h' '*.cpp' | xargs sed -i -e 's/^\s*#ifdef \(WORDS_NEED_ALIGNMENT\)\>/#if DXX_\1/' -e 's/\s*#\(el\)\?if \(.*\)defined(\(WORDS_NEED_ALIGNMENT\))/#\1if \2DXX_\3/' -e 's/^\s*#ifndef \(WORDS_NEED_ALIGNMENT\)\>/#if !DXX_\1/'
Rename symbol WORDS_BIGENDIAN to DXX_WORDS_BIGENDIAN to show that it is a DXX
symbol, not one inherited from a library.
This is a mostly automated transform, but the changes to SConstruct were
manual.
git grep -wl WORDS_BIGENDIAN -- '*.h' '*.cpp' | xargs sed -i -e 's/^\s*#ifdef \(WORDS_BIGENDIAN\)\>/#if DXX_\1/' -e 's/\s*#\(el\)\?if \(.*\)defined(\(WORDS_BIGENDIAN\))/#\1if \2DXX_\3/' -e 's/^\s*#ifndef \(WORDS_BIGENDIAN\)\>/#if !DXX_\1/'
Test DXX_HAVE_CONSTEXPR_UNION_CONSTRUCTOR probes for a bug present in
<=gcc-4.7.x. Rebirth now rejects <=gcc-4.8.x for other reasons, so all
supported gcc versions pass test DXX_HAVE_CONSTEXPR_UNION_CONSTRUCTOR.
Remove the test.
Rename symbol MAX_HATS_PER_JOYSTICK to DXX_MAX_HATS_PER_JOYSTICK
to show that it is a DXX symbol, not one inherited from a library.
git grep -lzw MAX_HATS_PER_JOYSTICK -- SConstruct '*.h' '*.cpp' | xargs -0 sed -i -e 's/\<MAX_HATS_PER_JOYSTICK\>/DXX_&/g'
Rename symbol MAX_BUTTONS_PER_JOYSTICK to DXX_MAX_BUTTONS_PER_JOYSTICK
to show that it is a DXX symbol, not one inherited from a library.
git grep -lzw MAX_BUTTONS_PER_JOYSTICK -- SConstruct '*.h' '*.cpp' | xargs -0 sed -i -e 's/\<MAX_BUTTONS_PER_JOYSTICK\>/DXX_&/g'
Rename symbol MAX_AXES_PER_JOYSTICK to DXX_MAX_AXES_PER_JOYSTICK to show
that it is a DXX symbol, not one inherited from a library.
git grep -lzw MAX_AXES_PER_JOYSTICK -- SConstruct '*.h' '*.cpp' | xargs -0 sed -i -e 's/\<MAX_AXES_PER_JOYSTICK\>/DXX_&/g'
Rename symbol MAX_JOYSTICKS to DXX_MAX_JOYSTICKS to show that it is a DXX
symbol, not one inherited from a library.
git grep -lzw MAX_JOYSTICKS -- SConstruct '*.h' '*.cpp' | xargs -0 sed -i -e 's/\<MAX_JOYSTICKS\>/DXX_&/g'
Rename symbol OGL to DXX_USE_OGL to show that it is a DXX
symbol, not one inherited from a library. Move it to dxxsconf.h to
shorten the command line.
This is a mostly automated transform, but the changes to SConstruct were
manual.
git grep -lzw OGL -- '*.h' '*.cpp' | xargs -0 sed -i -e 's/\(\s*#\s*if\)def\s*OGL/\1 DXX_USE_OGL/' -e 's/\(\s*#\s*if\)ndef OGL/\1 !DXX_USE_OGL/' -e 's/\(\s*#\s*if !\?\)defined(OGL)/\1DXX_USE_OGL/'
Rename symbol OGLES to DXX_USE_OGLES to show that it is a DXX
symbol, not one inherited from a library. Move it to dxxsconf.h to
shorten the command line.
This is a mostly automated transform, but the changes to SConstruct were
manual.
git grep -lzw OGLES -- '*.h' '*.cpp' | xargs -0 sed -i -e 's/\(\s*#\s*if\)def\s*OGLES/\1 DXX_USE_OGLES/' -e 's/\(\s*#\s*if\)ndef OGLES/\1 !DXX_USE_OGLES/'
Previously this could be achieved by setting the install path to @executable_path/../Frameworks/<path to framework> for each framework or library, but for Mac OS X 10.5 onwards, these frameworks (namely SDL and SDL_mixer) are set by default to install to @rpath for greater flexibility. @rpath is then set by the binary, in this case DXX-Rebirth to @loader_path/../Frameworks via SCons.
On Mac OS X it is common practice to distribute applications with any frameworks or libraries specific to that application within the application bundle.
clang on OS X incorrectly diagnoses mistakes in Apple system headers
even when -Wsystem-headers is not specified. Move -Wold-style-cast from
the forced CXXFLAGS section to an SConf test that includes one of the
deficient headers. This should cause SConf to declare OS X clang as
unable to use -Wold-style-cast, which is necessary to avoid spurious
failures when including Apple headers that use old style casts.
The legacy UDP tracker does not need either curl or jsoncpp.
The new HTTP tracker requires both. Add tests for both, guarded by
`use_tracker`. Force `use_tracker` to False for now. When the HTTP
tracker is made active, remove the assignment so that use_tracker=1 will
enable the new tests.
Rename symbol EDITOR to DXX_USE_EDITOR to show that it is a DXX
symbol, not one inherited from a library. Move it to dxxsconf.h to
shorten the command line.
This is a mostly automated transform, but the changes to SConstruct were
manual.
git grep -wl EDITOR -- '*.h' '*.cpp' | xargs sed -i -e 's/^\s*#ifdef \(EDITOR\)\>/#if DXX_USE_\1/' -e 's/\s*#\(el\)\?if \(.*\)defined(\(EDITOR\))/#\1if \2DXX_USE_\3/' -e 's/^\s*#ifndef \(EDITOR\)\>/#if !DXX_USE_\1/'
Rename symbol IPv6 to DXX_USE_IPv6 to show that it is a DXX
symbol, not one inherited from a library. Move it to dxxsconf.h to
shorten the command line.
This is a mostly automated transform, but the changes to SConstruct were
manual.
git grep -wl IPv6 -- '*.h' '*.cpp' | xargs sed -i -e 's/^\s*#ifdef \(IPv6\)$/#if DXX_USE_\1/' -e 's/\s*#\(el\)\?if \(.*\)defined(\(IPv6\))/#\1if \2DXX_USE_\3/'
Rename symbol USE_UDP to DXX_USE_UDP to show that it is a DXX
symbol, not one inherited from a library. Move it to dxxsconf.h to
shorten the command line.
This is a mostly automated transform, but the changes to SConstruct were
manual.
git grep -l USE_UDP -- '*.h' '*.cpp' | xargs sed -i -e 's/^\s*#ifdef \(USE_UDP\)$/#if DXX_\1/' -e 's/\s*#\(el\)\?if \(.*\)defined(\(USE_UDP\))/#\1if \2DXX_\3/'
Rename symbol USE_TRACKER to DXX_USE_TRACKER to show that it is a DXX
symbol, not one inherited from a library. Move it to dxxsconf.h to
shorten the command line.
This is a mostly automated transform, but the changes to SConstruct were
manual.
git grep -l USE_TRACKER -- '*.h' '*.cpp' | xargs sed -i -e 's/^#ifdef \(USE_TRACKER\)$/#if DXX_\1/' -e 's/#\(el\)\?if \(.*\)defined(\(USE_TRACKER\))/#\1if \2DXX_\3/'
Rename symbol USE_SDLMIXER to DXX_USE_SDLMIXER to show that it is a DXX
symbol, not one inherited from a library. Move it to dxxsconf.h to
shorten the command line.
This is a mostly automated transform, but the changes to SConstruct and
inferno.cpp were manual.
git grep -l USE_SDLMIXER -- '*.h' '*.cpp' | xargs sed -i -e 's/^#ifdef \(USE_SDLMIXER\)$/#if DXX_\1/' -e 's/#\(el\)\?if \(.*\)defined(\(USE_SDLMIXER\))/#\1if \2DXX_\3/'
clang does not understand -Wuseless-cast. Passing -Wuseless-cast to
clang causes clang to fail the compile. Resolve this by adding an SConf
test for -Wuseless-cast instead of adding it unconditionally. It cannot
be put in the compiler preferred options tuple since it must be set
before the test for whether casting a ptrdiff to int is useless.
The folded form combined multiple targets into one, which is not the
desired result.
'''
unfolded
l=['gcc5,', 'mingw5,'] e='gcc5,' p=('gcc5', '')
l=['gcc5,', 'mingw5,'] e='mingw5,' p=('mingw5', '')
folded
l=['gcc5,', 'mingw5,'] e='mingw5,' p=('gcc5', '', 'mingw5', '')
'''
import itertools
l = ['gcc5,', 'mingw5,']
print('unfolded')
for e in l:
for p in itertools.product(*[v.split('+') for v in e.split(',')]):
print("l=%r e=%r p=%r" % (l, e, p))
print('folded')
for p in itertools.product(*[
v.split('+') \
for e in l \
for v in e.split(',') \
]):
print("l=%r e=%r p=%r" % (l, e, p))
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.
Newer SCons applied a performance optimization that breaks a technique
used in the check_header_includes=1 build. Switch to a technique that
should work reliably on old and new SCons, although with slightly
reduced efficiency in both cases. The effective performance penalty
should be unnoticeable and only applies to developers building with
check_header_includes=1.
The C++11 required feature string was computed once per configured
environment. When two or more environments were configured in a single
run, this was wasteful. Reorder the SConf tests to precompute the
string.
Commit 957016621e ("Remove support for
Boost.TypeTraits") removed support for any alternative implementation of
<type_traits>, so C++11 <type_traits> is now required. Move it to
the Cxx11RequiredFeature array to combine it with the other required
feature tests.
Fallback support for Boost.Foreach is broken and no one has reported it.
The minimum required gcc version supports range-based for. Recent Clang
supports range-based for. No Microsoft compilers are supported. Remove
support for the fallback Boost.Foreach and move the C++11 range-based
for test to the Cxx11RequiredFeature array.
Building with Boost.Array fails due to name lookup errors. These could
be fixed, but since no one has reported them and they are fairly old,
Boost.Array appears to be unused. Remove support for it.
Building with std::tr1::array fails due to missing features in
std::tr1:array relative to std::array. No one has reported this either,
so remove support for std::tr1:array.
Move the test for std::array into the Cxx11RequiredFeature list so that
it is run as part of the group test, rather than as a separate
statement.
GCC std::array uses std::size_t for its size_type. On Linux/amd64,
`std::size_t` is `unsigned long`. On Win32, `std::size_t` is
`unsigned int`. This provokes format string warnings because an
unsigned int is passed where the format string declares an unsigned
long. On Win32, `unsigned int` and `unsigned long` are the same size,
so `unsigned long` could have been used to avoid this problem by
maintaining consistency with Linux, but it was not. This may have been
an attempt to achieve bug compatibility with Microsoft's types.
Add a workaround by defining a macro DXX_PRI_size_type in the style of
inttypes.h PRI* macros. Use an SConf test to determine the correct
value by inspecting which compilation runs succeed. Currently, Linux
needs "l", Win32 needs "", and Win64 needs "I64".
Some tests may benefit from the results of earlier tests. Other tests
may give different results if the macros from earlier tests are defined.
SCons dependency checking rejects including dxxsconf.h during testing,
so save the results of self._check_macro and insert them into later
tests. Some tests define their macros in other ways. The macros from
those tests and are not yet inserted.
Boost enable_if works differently from std::enable_if, so the build is
broken when using Boost.TypeTraits. The broken code has been present
for a long time and no one reported it. Remove the fallback to Boost.
Windows 2000 has no support for getaddrinfo. Five years ago (Apr 5
2011), the addition of tracker support
(7f18afd5d7) brought a requirement for
getaddrinfo. Recently, someone with a working Windows 2000 system
reported that Rebirth no longer starts on Windows 2000. Although such
systems should not be connected to the Internet, they may still need
name resolution on a LAN. Add a configure test to probe for getaddrinfo
support and fall back to gethostbyname if needed.
Compiling a binary that works on Windows 2000 requires one of:
* using an ancient SDK that does not declare getaddrinfo
* defining _WIN32_WINNT to exclude Windows XP support
* overriding the SConf test to force failure
Fixes: 7f18afd5d7 ("Client-side implementation for Tracker support by Matt 1360 Vandermeulen including improvements in udp_dns_filladdr and IPv4/IPv6 compability; Very little adjustments by me, too including IPv6 support for Windows (untested); Actual tracker code will follow later as seperate branch when it's done")
Clang becomes confused trying to resolve X::a when:
- both `namespace ::X` and `class ::B::X` exist
- `using namespace B;` is active
even if class ::B::X is only a forward-declaration. Add an SConf test
to recognize when the compiler rejects this pattern and hide the safety
nets that are implemented using it.
Microsoft cmd.exe has strange quoting rules. Where other shells treat
single quote as a quote character, cmd.exe seems to pass it through to
the underlying process. This breaks the vers_id compilation with
strange error messages. In practice, SCons does not need the
no-interpolation semantics that single quote provides, so switch to
double quote for compatibility with Windows. This is probably slightly
less efficient on Bash, since it will now scan the value for expansions.
However, the efficiency loss is tiny compared to overhead elsewhere in
the build.
Per the documentation, on Linux, the default is true. On Windows,
setting it to true causes a Python exception because Windows relies on
disabling inheritance as a hack to close all descriptors, but
inheritance needs to be enabled to capture the output of the child
process.
Reported-by: Mako88 <https://github.com/dxx-rebirth/dxx-rebirth/issues/173>
Fixes: bb3f414128 ("Cache scons subprocess calls")
Move the boilerplate up onto the outer namespace line. Decorate each
collision with a comment stating its purpose, so that compiler error
output echoes the type of collision.
This started as a project to make SConstruct take less time on an empty
build. Unforunately, most of the major gains lie elsewhere and require
deeper changes to the build system. These changes do not help much, but
they are better than nothing.
Add option to compute indirect includes even when not necessary, since
indirect includes influence the inclusion count shown in the comments.
Enabling this option makes SCons work harder for the same output binary.
It should only be used by developers who want to see precise file
inclusion counts.
Current gcc requires a non-PIE cc1plus for working PCH support. If the
chosen compiler is PIE, it will succeed in building the PCH, then fail
when trying to use it.
Per issue from MattWatt, gcc for Raspberry Pi warns about alignment
issues.
Per issue from kreator, clang warns about alignment issues on OS X.
However, it does not warn on x86_64-pc-linux-gnu.
GCC only warns when compiling for a target architecture that requires
alignment, so it does not warn for i686-pc-linux-gnu or
x86_64-pc-linux-gnu.
Using this warning would be nice, but since it only triggers on
architectures not maintained by active developers, it causes more
trouble than it may solve.
Alignment sensitive users will have to get by with software fixups until
an active developer begins targeting alignment sensitive hardware.
Reported-by: MattWatt <https://github.com/dxx-rebirth/dxx-rebirth/issues/119>
Reported-by: kreatordxx <https://github.com/dxx-rebirth/dxx-rebirth/issues/120>
Some builds of SDL lack joystick support. Rebirth requires joystick
support in SDL, even if the user never connects a joystick. Add a call
to SDL_NumJoysticks during the SDL test so that systems without SDL
joystick support fail the "usable library SDL" test.
Test check_compiler_template_parentheses_warning probes for a bug in
gcc-4.6. Upstream fixed the bug in gcc-4.7. Using gcc-4.6 is no longer
supported, so remove the test for the bug.
clang does not support -flto=N syntax for parallel LTO.
Remove -fno-fat-lto-objects too. clang does not understand it;
gcc 4.9 and later default to non-fat LTO.
Fixes: 1037a92d60 ("Allow parallel lto")
Previously, SConstruct tested whether static_assert worked with trivial
constant expressions. Extend the test to more complicated expressions
like the ones used in the program.
Replace the typedef based approach with an inline enum to avoid warnings
about unused local typedefs.
Allow use with compilers that lack working static_assert or that cannot
accept complicated expressions. Add a non-checking definition that uses
the input parameter. This prevents warnings when types or constants are
used only for the static_assert.