Heapify closures on OSX/FreeBSD.
Add thread-local storage for POSIX-y platforms.
Fix boundary-spanning delimiters.
Don't readinto if we're dropping.
Actually read delimiter length bytes. Don't index out of bounds.
Fix missing free in aesgcm (thanks mike)
Try /dev/urandom before /dev/random.
Unswap sha output bytes. (Thanks Mike!)
Fix 'make clean' (Thanks Mike)
Add iterator for all section, key, val pairs in inifile.
Wait can't always be called from other procs.
Implement futures + thread.do
Add -v option, make it print test output.
Only attempt to ftxwake in sempost if there might be a waiter.
Add rwlocks.
Fix futex timeouts and handle futex error codes.
Make timespec/timeval struct members signed to simplify arithmetic.
Clean up draining incqueues.
Revert "Fix overlapping match statements (thanks Mako)"
Make timespec/timeval struct members signed to simplify arithmetic.
We don't need early breaks.
Clean up the mutex code a bit more.
We can pick fma generically at compile time.
We don't need the impls in sin-impl.
Record the closure into all nested funcs.
Prettify mutex code a bit.
Merge remote-tracking branch 'npnth/libmath'
Include test cases for all (relevant) sub-polynomials of atan.
Correct extra-precision division typo for atan.
Correctly compute u, du for y/x in atan.
Improve accuracy of poly path of atan calculations.
Negate atan properly when atan(1/x) is computed.
Allocate condvar waiters on heap.
Fix typo in plan 9 mksem.
Subject: [PATCH 2/2] Add/fix condvar implementations.
Add ftxwakeall and waitgrps to libthread.
Adjust leeway in arctan tuple generator.
Document that generate-minimax-by-Remez.gp has issues.
Add tuple-generation for arctan.
Don't abort when we can't open a file.
Handle awkward special cases in tan/cot32.
Test Highly Accurate Tables for tan() and cot().
Handle -0.0 correctly in tan() and cot().
Correct typo in cotangent calculations.
Add zeroed env pointer when a const fn is passed as an argument.
Fix codegen bug in env capture.
First compiling tan/cot functions.
Add documentation for tuple access expressions
Support direct tuple access operators "tuple.N"
Top level anonymous unions may not have their type set.
Update triple-generate to also compute tables for tan and cot.
Push down named unions types into union tags
Make bio and regex disposable.
Fix typo in generic sleq implementation
Make bio and regex disposable.
Add match tree debug option.
Enable a few more tests that we fixed.
Fix overlapping match statements (thanks Mako)
Use `eq()` function in generic equatable slices.
Fix auto expressions for complex values.
Typo fix in libregex.
Fix infinite recursion in type inference.
Respect {noinst} attribute.
Fix escaping of closing '}' in fmt.
Add nopipe function to disable sigpipe.
Make '!=' not hard coded as evaluating to '==' for enum unions.
Drain increment queues before jumping in matches.
Fix build on OSX: Export missing sys.cstring function
Document results of sin/cos a bit more.
Eliminate loop sorts in sin/cos reduction.
Cut down results of reduce() from triple to double for sin/cos.
Remove a sort or two in sin/cos.
Reduce overkill on precision in a few sum computations for sin/cos.
Add test cases for each C[j] constant for sin/cos.
Be a little smarter about downscaling flt64s to flt32s in sin/cos.
Find nearest xi value, instead of blindly computing j.
Fix local reduction from [-pi/2, pi/2] to [-pi/4, pi/4]
Rewrite xa, xb, xc calculation in huge_reduce.
Initialize variables (fallout from upstream merge?).
Merge remote-tracking branch 'ori/master' into sin-cos-work
Fix bi/ci calculation in reduction.
Add macros for relative error to lib/math/ancillary.
Initial (only compile-tested) implementation of sin, cos.
Apply the same bootstrap changes to 9front.
Make bootstrap more seamless.