continuations work, more assembly instructions are testing in 64-bit
mode, and some arithmetic operations (shift-right, etc.) have been
fixed by removing some 32-bit dependencies.
- (optimize-level [0,1,2]) and ikarus -O[0,1,2]
where -O0 = no optimizations
-O1 = using old optimizer
-O2 = using the new cp0 optimizer
defaults to -O1 for now.
- (cp0-size-limit n) which is the limit of the residual size for
each inlining attempt
- (cp0-effort-limit n) which is the limit on the effort expended
for each inlining attempt
- Rewrote the syntax-match macro to make use of the same technology
used in syntax-case itself resulting in reduced code size.
- Added (system-value <symbol>) which returns the system value.
E.g., (system-value 'car) => #<procedure car>
This is pretty much the same as
(eval <symbol> (environment '(ikarus)))
except that it does not involve compiling the expression or
consulting the library/expander systems.
- Fixed the fasl loader to make it understand complex numbers.
- Added file-regular?, file-directory?, file-symbolic-link?,
directory-list, make-directory, delete-directory, change-mode,
and make-symbolic-link.
- Bug fix in ikrt_open_input_fd and ikrt_open_output_fd which
considered a 0 FD as an error.
- Bug fix in ikrt_getcwd about len+1.
- Bug fix of &i/o-file-protection which didn't have &i/o-filename as
its parent.
- There is a new src/ikarus-errno.c file.
- Made tcp-connect and friends tell when they "failed to resolve
host name or connect".
- To support the uses of &i/o-filename subtypes, I modified
print-condition so it will cleanly print the parents' fields of a
condition;
read or write operation would block. Instead, they are schedules
with an event handler that `selects' on the pending file descriptors
and dispatches the appropriate callback.
manually inserted into the code stream to ensure proper offsets.
These need to be removed ASAP. Also, abstracted the values of three
constants that were hardcoded.
socket in nonblocking mode. An operation that would block now
raises (continuable) a condition of type &i/o-would-block which
contains the port in question. There's no way to handle the
condition gracefully yet.
name (e.g. "www.google.com" and "http") and returns two binary ports:
the first for output and the second for input.
Both ports must be closed for the connection to close.
Also added an example in lab/tcp-connect-example.ss which connects
to an http server, sends "GET /\n" and prints the responde.
args to it, and returns 4 values:
* the process's pid
* the process's standard-input-port (for writing)
* the process's standard-output-port (for reading)
* the process's standard-error-port (for reading)
See lab/process-example.ss for a sample usage.
Previously: when a generation (say 2) is collected,
everything in gen 0 moves to gen 1
everything in gen 1 moves to gen 2
everything in gen 2 moves to gen 3
Now: when a generation (say 2 again) is collected
everything in gen 0 moves to gen 3
everything in gen 1 moves to gen 3
everything in gen 2 moves to gen 3
So, some objects get a leap from young to old if they're lucky to be
in the right time.
Consequences: when an object is moved by the collector, we don't
need to track old->new pointers and masks because all moved objects
are clean now. This both simplifies the collector and makes it more
efficient and might open the door for further optimization
opportunities. For bootstrap time, we get about 5% overall
saving and about 20% GC-time saving. Not bad.
BEFORE:
running stats for macro expansion:
45 collections
2558 ms elapsed cpu time, including 212 ms collecting
2576 ms elapsed real time, including 216 ms collecting
186972152 bytes allocated
running stats for code generation and serialization:
86 collections
4365 ms elapsed cpu time, including 1444 ms collecting
4374 ms elapsed real time, including 1449 ms collecting
362819096 bytes allocated
running stats for the entire bootstrap process:
131 collections
6928 ms elapsed cpu time, including 1657 ms collecting
6953 ms elapsed real time, including 1666 ms collecting
549818232 bytes allocated
AFTER:
running stats for macro expansion:
45 collections
2506 ms elapsed cpu time, including 169 ms collecting
2511 ms elapsed real time, including 171 ms collecting
186968056 bytes allocated
running stats for code generation and serialization:
86 collections
4083 ms elapsed cpu time, including 1189 ms collecting
4085 ms elapsed real time, including 1191 ms collecting
362810904 bytes allocated
running stats for the entire bootstrap process:
131 collections
6591 ms elapsed cpu time, including 1359 ms collecting
6599 ms elapsed real time, including 1362 ms collecting
549805944 bytes allocated
Happy Happy Joy Joy
Abdulaziz Ghuloum