ObjFW  History of src/OFThread.h of 2cb48ceaf1d0c54a929080b17bcf0f994335809830ae6d0cbd8ae81dc9f2644b

__COUNTER__ does not exist in GCC 4.2, and Apple GCC 4.2 is still the
newest compiler available for macOS 10.5. file: [e49332cd5c] check-in: [ef614a225d] user: js, branch: trunk, size: 7902 [annotate] [blame] [check-ins using] [diff]

Getting a pointer and increasing it until we reach the top pointer does
not work: Releasing an object can temporarily create new autorelease
pools, which can trigger resizing of "objects" using realloc, which can
move it to a different address, which will then lead to continuing to
iterate on a now invalid pointer.

This is now solved by using an index into "objects" instead. Since we're
now indexing for the pop, let's use indexes everywhere, as they're more
readable anyway.

While debugging this, I noticed that the last pool is popped quite
frequently, only for a new pool to be pushed immediately again. This
resulted in a free followed by a malloc every time. Instead, keep the
pool, but let OFThread explicitly say when to free everything. file: [4c0a8350b3] check-in: [e7f4f80e23] user: js, branch: trunk, size: 7892 [annotate] [blame] [check-ins using] [diff]

This allows ABI stability with the fragile ABI. file: [1d8bb7f623] check-in: [496695d778] user: js, branch: trunk, size: 7898 [annotate] [blame] [check-ins using] [diff]

This allows the runtime to only link against the parts it needs, without
pulling in unnecessary parts like thread spawning, TLS and conditions. file: [dc390a6e23] check-in: [5358e9ea6a] user: js, branch: trunk, size: 7095 [annotate] [blame] [check-ins using] [diff]

Some systems do not provide the functionality. file: [deb7104407] check-in: [b5430ac44f] user: js, branch: trunk, size: 7098 [annotate] [blame] [check-ins using] [diff]

This returns the DNS resolver for the current thread. file: [76cb344740] check-in: [2fb0769744] user: js, branch: trunk, size: 7052 [annotate] [blame] [check-ins using] [diff]

This improves Swift compatibility. file: [8db864191c] check-in: [a2e849c681] user: js, branch: trunk, size: 6554 [annotate] [blame] [check-ins using] [diff]

This improves Swift compatibility. file: [a03e0a91d4] check-in: [f3ccf14461] user: js, branch: trunk, size: 6187 [annotate] [blame] [check-ins using] [diff]

Instead, explicitly declare them, as OF_ASSUME_NONNULL_{BEGIN,END} does
not apply to implicit return types. This means that after this commit,
all init methods have a nonnull return type, as they should have. file: [42f39cebaf] check-in: [2f4e0df8be] user: js, branch: trunk, size: 6832 [annotate] [blame] [check-ins using] [diff]

The name of the thread may only be set from the thread itself. So the
thread should read the property on start and set its name. file: [c358f93979] check-in: [1494d3f6f1] user: js, branch: trunk, size: 6153 [annotate] [blame] [check-ins using] [diff]

Casts are now written like types in variable declarations. file: [04bcc49824] check-in: [4af49a13c3] user: js, branch: trunk, size: 6426 [annotate] [blame] [check-ins using] [diff]

This changes retaining behavior, meaning properties are not returned
retained and autoreleased anymore, so a property returned from a getter
now needs to be manually retained and autoreleased before calling the
setter. However, this is rarely the case and not using atomic improves
performance. file: [5cd28cdb35] check-in: [0a73af49f0] user: js, branch: trunk, size: 6419 [annotate] [blame] [check-ins using] [diff]

Forgot to add 2017, even though I already did quite some changes in
2017. file: [4c510e12e2] check-in: [44f45c2e35] user: js, branch: trunk, size: 6386 [annotate] [blame] [check-ins using] [diff]

The old definition was quite unpredictable and too platform-specific.
The new one defines -1.0 as lowest priority that still schedules, 0.0 as
the normal priority (meaning the same as the main thread) and +1.0 as
the highest priority that still allows getting preempted. file: [383f3c4912] check-in: [b4023e6bc0] user: js, branch: trunk, size: 6380 [annotate] [blame] [check-ins using] [diff]

While at it, also update the mail address. file: [3ac85a8b66] check-in: [cec0f072f8] user: js, branch: 0.8, size: 6011 [annotate] [blame] [check-ins using] [diff]

While at it, also update the mail address. file: [7293edf304] check-in: [2a27cf3000] user: js, branch: trunk, size: 5944 [annotate] [blame] [check-ins using] [diff]

This increases the required GCC version from 4.0 to 4.6 (exception:
Apple GCC, which already supports this with >= 4.0 starting with OS X
10.5). Since even GCC 4.6 is really old by now, there is no point in
still supporting something even older and making the code ugly because
of that. While some hardware and OS support was dropped from GCC 4.6
compared to GCC 4.0, there is nothing in there that would be an
interesting target with the exception of BeOS maybe - but a port to BeOS
can also be achieved using the Haiku support. The other dropped OSes are
mostly old versions of OSes while newer ones are still being supported
(and those newer versions of those OSes still support the same
hardware). file: [cd0fc73b17] check-in: [48980f2297] user: js, branch: trunk, size: 5940 [annotate] [blame] [check-ins using] [diff]

Now that Clang switched to use _Nullable and _Nonnull instead of
__nullable / __nonnull, there is no longer a conflict with glibc, which
means we can just define _Nullable / _Nonnull to nothing if they are not
understood by the compiler (which did not work with __nullable /
__nonnull due to this conflict).

This also defines _Null_unspecified to nothing if unavailable. file: [5636f9d458] check-in: [d526d938d7] user: js, branch: trunk, size: 6005 [annotate] [blame] [check-ins using] [diff]

Using __nonnull directly doesn't work on systems using glibc, as glibc
defines __nonnull as a parameterized define. While this does not fix the
problem of Clang introducing __nonnull even though it conflicts with
glibc, this at least means it's possible again to compile things with
versions of Clang that don't support __nonnull on systems with glibc. file: [81d5aa4cdd] check-in: [b7097a67b6] user: js, branch: trunk, size: 6007 [annotate] [blame] [check-ins using] [diff]

If it's unavailable, it's defined to __attribute__((noreturn)).

Unfortunately, it cannot be used for ObjC methods, as noreturn is part
of the return type while __attribute__((noreturn)) needs to be at the
end for an ObjC method. To make matters worse, even GCC versions that
accept noreturn don't allow it for an ObjC method. Thus, the only thing
that can be done is to always use __attribute__((noreturn)) for ObjC
methods using the OF_METHOD_NORETURN define. file: [4ccbd9a433] check-in: [bac91ccede] user: js, branch: trunk, size: 5097 [annotate] [blame] [check-ins using] [diff]

This is used instead for time intervals now instead of doubles. file: [42d02c0030] check-in: [dc31ce3d34] user: js, branch: trunk, size: 5009 [annotate] [blame] [check-ins using] [diff]

It is now called -[initWithThreadBlock:] in order to prevent method
signature conflicts.

This also fixes a missing release for the thread block. file: [a1b30840b5] check-in: [a3639589d3] user: js, branch: trunk, size: 5571 [annotate] [blame] [check-ins using] [diff]

This means that even if --disable-threads has been specified, it is
possible to use +[OFThread sleep*] to sleep the current (and only)
thread. file: [b4c8773e10] check-in: [71d45a29d1] user: js, branch: trunk, size: 5523 [annotate] [blame] [check-ins using] [diff]

This also removes "bool *stop" from some blocks where it does not make
sense. file: [b48c7d5a0b] check-in: [e8473b9db8] user: js, branch: trunk, size: 5380 [annotate] [blame] [check-ins using] [diff]

On Haiku, this even renames the Haiku thread so that in a process
manager the thread's name can be seen. If no name has been set, it uses
the class name as name for the Haiku thread. file: [b43535f12f] check-in: [debba8efb6] user: js, branch: trunk, size: 5217 [annotate] [blame] [check-ins using] [diff]

Blocks don't allow passing it anymore and without blocks, subclassing is
necessary anyway so that an ivar with the correct type can be added
(instead of id) if required. file: [9008bb5cf4] check-in: [0639a351db] user: js, branch: trunk, size: 4881 [annotate] [blame] [check-ins using] [diff]

It was time to finally have one file per class. file: [384da59e46] check-in: [0a55edad35] user: js, branch: trunk, size: 5262 [annotate] [blame] [check-ins using] [diff]

Newer clang versions don't allow the retain attribute for blocks
anymore. file: [2dc62c0de7] check-in: [4d18a4b596] user: js, branch: trunk, size: 7294 [annotate] [blame] [check-ins using] [diff]

Also fixes the forgotten change to double in +[sleepForTimeInterval:]
and fixes two warnings due to missing casts from double to uint32_t. file: [f4f9d56218] check-in: [c19b2b5726] user: js, branch: trunk, size: 7290 [annotate] [blame] [check-ins using] [diff]

We assume now that every object understands retain, release, etc. so
that we can weaken the type from OFObject* to id. This makes it
possible to use different root object classes. file: [cc3937ad7a] check-in: [e2b06423e0] user: js, branch: trunk, size: 4134 [annotate] [blame] [check-ins using] [diff]