| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| | |
|
| |
|
|
| |
Fix CI by disabling non-native runs that run out of memory.
|
| |
|
|
| |
Include phases of self-hosted SubX translator in CI.
|
| |
|
|
| |
Include simple apps in CI.
|
| | |
|
| |
|
|
| |
Fix a timeout in CI.
|
| |
|
|
|
| |
We now have a new pass called 'tests' which code-generates a new
function called 'run-tests', just like the C++ layer `tests.cc`.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
| |
Bugfix fourteen: we need different address computation logic for code vs
data labels.
It's really about different categories of instructions having different
address computation logic. This subtle distinction will make good error
messages hard. But that's a problem for later.
Now there's just one example program not translating.
|
| |
|
|
|
|
|
|
| |
Bugfix thirteen: displacement calculations were wrong because current
offset was not being updated properly as words were being read and
emitted.
Now 10/12 example programs are translated correctly.
|
| |
|
|
| |
Fix CI.
|
| |
|
|
| |
Fix CI.
|
| |
|
|
|
| |
Clean up other examples as well to satisfy the requirements in commit
5404.
|
| |
|
|
|
| |
Bugfix eleven: segment flags were incorrectly computed. examples/ex1 now
verified! Added to CI.
|
| | |
|
| |
|
|
| |
Fix CI for a real, restricted /bin/sh.
|
| |
|
|
|
| |
After fixing bugs, checking alloc_ids now requires 9 instructions rather
than 6.
|
| |
|
|
| |
Add the new binary to CI.
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
| |
Pull in a _different_ function than `next-word` (commit 5092) into a shared
file between phases. Let's see how this goes.
|
| |
|
|
|
|
|
| |
Realization: 'next-word' can't be reused in converting string literals,
because it has to understand string literals.
Let's just keep each phase self-contained.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Current plan for SubX translator:
$ cat files.subx ... |assort |pack |survey |hex > a.out
Higher-level notations will be inserted at the start of the pipeline. The
first (and needed for bootstrapping) is for string literals.
$ cat files.subx ... |string-literals |assort |pack |survey |hex > a.out
Alternatively, we should check how often we use string literals and just
convert them by hand.
They're used all over in tests, and converting them would make tests hard
(even harder) to read.
|
| | |
|
| |
|
|
|
| |
Build the C++ version optimized by default when building/running all apps.
We have enough apps now that the cost of optimized builds is worthwhile.
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Done with kinda-safe pointers.
In a real compiler the fast path of 'lookup' would ideally get inlined.
Excluding procedure-call overhead, the current implementation consumes 2
registers besides the input, and requires 9 instructions (2 push, 2 load,
compare, jump, increment, 2 pop). That's large enough that inlining may
become a trade-off. Even if we somehow magically had the registers already
loaded and available, we'd still need 4 instructions (1 pointer dereference,
compare, jump and increment). The price of safety.
|
| |
|
|
| |
We only can't use rm32=5 when mod=0. Totally fine when it's mod=1.
|
| | |
|
| |
|
|
|
|
|
| |
Start with an exactly corresponding version to Crenshaw 2-1: single-digit
numbers. The only change: we assume the number is in hex.
The next version now supports multi-digit hex numbers.
|
| |
|
|
| |
Simplification.
|
| |
|
|
|
| |
This time I've ported (and test-driven) 'GetChar' and 'GetNum'. The new
tests bring together our new testable interfaces for read() and exit().
|
| | |
|
| |
|
|
|
|
| |
Start injecting all dependencies in the Crenshaw compiler app.
In the process I realized the non-fake code path of 'stop' had a bug.
|
| |
|
|
|
|
|
| |
I've started testing more regularly on 64-bit Linux, and I realize that
the binaries actually now compare identical even on 64-bit. Why did I ever
add that condition to CI? Seems to be working now on Travis. Let's start
comparing now and see if the problem happens again.
|
| |
|
|
|
|
|
|
|
|
| |
Turns out the tests for 'trace' have been broken in native mode since the
original commit (4674). Dangers of running my tests on Darwin, where I
can't run them natively.
The test failures didn't get flagged on CI because I'd forgotten to update
the exit code of the factorial app in commit 4664. At least that's fixed
in this commit.
|
| | |
|
| | |
|
|
|
Generated 32-bit binaries are different on 64-bit.
So let's compare them only on a 32-bit platform.
And let's start also verifying their run-time behavior on Linux.
|