2d835cffa6
Fix the lock order when sending +initialize, so that we don't hold a lock that
...
is meant to protect a runtime data structure while trying to acquire a lock
that must be held when running user code that may take an indeterminate amount
of time and may trigger deadlocks.
14 years ago
855f2601c9
Fix some corner cases where exceptions are thrown through +initialize.
14 years ago
27452f82f0
Some bug fixes.
14 years ago
26d86e101c
Inherit +initialize methods.
15 years ago
8af52846b7
Small object (hidden in a pointer) support.
15 years ago
7b6ba21ce5
Small bug fixes:
...
- Don't call C++ constructors if they don't exist.
- Don't check the owner of retain / release methods if they are not implemented.
- Add arc.m as a file to ignore when checking for GC compatibility.
15 years ago
e34be81404
Fix fast-path test to work with subclasses correctly.
15 years ago
34f7baf8d7
Added a fast path for ARC. Now, if a class implements ARC-compatible retain / release / autorelease methods, we don't call them at all. Instead, we inline them in the ARC accessors. This avoids all of the overhead of the message send (lookup and call) and should make ARC quite a bit faster than manual reference counting.
15 years ago
ad16172625
First pass at implementing Apple-compatible GC using Boehm. Still needs some tidying, but the following work:
...
- __strong pointers, preventing objects from being freed
_ __weak pointers are automatically freed when the last remaining __strong pointer goes away
- objc_gc_{retain,release}_np() functions, which can be used to implement CFRetain() and CFRelease() (adds an reference count - the object will not be collected until after its last retain is gone.
15 years ago
12385fcb1e
Tidy up low memory profile to use slots directly.
15 years ago
709f238364
Sync on the metaclass, not the class, since that's what we're actually testing...
15 years ago
f9d1b5f356
Clean up the low memory profile. Now passes all GNUstep-base and EtoileFoundation tests, just like the default profileand uses 5-10% less (total) RAM in GORM. Worth benchmarking to see how much performance this costs.
...
Updated the release announcement.
15 years ago
7b49755964
Dtable cleanup cleanups.
15 years ago
bad36445e3
Improve efficiency of dtable updates from categories.
...
Make +initialize sending exception safe.
15 years ago
987ab88a94
Don't do hidden-class transforms for classes, just dangle the look-aside data off the class structure.
...
Individually lock classes so that +initialize can be sent concurrently to two different classes in two different threads.
15 years ago
d51500184c
More cleanups. Stop using __ identifiers for internal stuff now that it's properly marked private and we don't have to worry about conflicts.
15 years ago
6c7cf4a5a6
Add visibility attributes on all internal functions so that we aren't ever exporting them outside of libobjc.
15 years ago
a2db6f58be
Made the selector name field into a union of the name (unregistered selectors) and the selector id / index (registered selectors), to more accurately reflect how it is used in the code.
15 years ago
e11e8dee46
Tweak ordering when creating dtables to fix an irritating corner case in +initialize calls.
15 years ago
7be9f34619
Fixed @synchronize() with a class that has had an instance used for @synchronize().
15 years ago
e9b129d878
Added low memory profile to libobjc2. This uses a more NeXT-style dtable design, which gives better memory usage at the cost of (significantly) worse worst-case performance. The memory savings seem to be only about 10% in the apps I've tested (Gorm, edlc), so it's probably not worthwhile, but it might be useful for someone finding stuff is just starting to swap...
16 years ago
121e14a3b8
Small cleanup of dtable code.
16 years ago
0e374db1db
Added support for type-dependent dispatch to libobjc2.
16 years ago
779b28abeb
Lots of tidying, removing legacy stuff.
16 years ago
b1b9baf382
Finished rewrite of message sending. Deleted lots of legacy stuff.
16 years ago
bb8cafaa15
Finished rewriting the loader. It's now cleanly separated into logically-separate components, so there's a chance it might actually be maintainable...
16 years ago
55a62b527f
Removed unused variable that prevents compilation by being treated as an error
16 years ago
8a451d212c
More dtable fixes - now the EtoileFoundation tests work when compiled with GCC too. Hopefully that's all of the irritating corner cases in category loading dealt with...
16 years ago
e482b6986f
More fixes for the new dtable. Now passing all of the EtoileFoundation tests, which do lots of things involving categories, so make a good stress test.
...
No longer installs headers by default. This is because GNUstep Make notices when the objc headers have changed (because GNUstep includes them) and so reinstalling libobjc2 was requiring a complete recompile of anything that you tried to build, even if you only changed one line. A better fix for this would be for install to use cmp or diff to check if the header has been modified before installing it, but I'm too lazy to do that right now.
16 years ago
7a9b105817
Fixed missing ! in dtable construction code, which meant that class methods were being preferred to category methods.
16 years ago
c40f9e84f2
Return NULL instead of "" from sel_getTypes_np() - matches behaviour of sel_get_types() and makes more sense.
16 years ago
1a60fabf96
More 64-bit fixes.
16 years ago
d36b939232
Actually commit the new dtable implementation this time...
16 years ago
theraven
qmathe