fernando/naravm

Watch

Fork 0

rework vm file format #7

Merged

fernando merged 68 commits from develop into main

2026-06-23 21:35:54 +00:00

fernando commented

2026-06-23 21:35:44 +00:00

Owner

No description provided.

fernando added 68 commits

2026-06-23 21:35:45 +00:00

feat: minimal deserialization cbe991cd64

test: deserial 6e9f01794d

feat: deserialization of bytecode w strings 2ae6324dbf

feat: use bytes from wasm input for the VM execution 447336541f

feat: run bytecode sent from JS 2f3d622b07

feat: serialize chunk 8546df4b85

Merge pull request 'feat/deserialization' (#1 ) from feat/deserialization into develop 82af16b3ac

Reviewed-on: fernando/thp-vm#1

feat: string concatenation b51c480678

feat: jumps ce60a29910

feat: negative jumps 9812deb9bc

feat: smoll changes 5e65cab012

Merge branch 'develop' into feat/deserialization 1c79f1ee72

Merge pull request 'feat/strings' (#3 ) from feat/strings into develop b475d404e0

Reviewed-on: fernando/thp-vm#3

Merge branch 'develop' into feat/deserialization 96bfc98f02

Merge pull request 'feat: smoll changes' (#2 ) from feat/deserialization into develop d8a63f4af5

Reviewed-on: fernando/thp-vm#2

feat: small changes b631b0139c

feat: read bytecode from stdin & dissasemble 93b6bea856

feat: 64 bit numbers & operations 079a5f8e4f

fix: i64 div b64d0580a6

feat: cast 64bit numbers to string 0bc6e0f30b

feat: op_cmp 124a7e12f8

fix: bool_to_s bug 1bd7ea2daf

chore: name change on package 0104319a46

chore: change package hash b5b088e0e0

chore: change package name c266995f69

feat: write i16 as big endian bytes 8d48a55027

feat: patch i16 in bytecode 48dab0cadf

feat: dissasemble jmpz 124d2fd126

feat: show jump address of jmpz on dissasemble e3a6777e3a

fix: jmpz dissasemble label 70b44dee3a

feat: jenkinsfile build

naravm/pipeline/head This commit looks good

Details

2c87674d8b

feat: dissasemble jmp 1c40e4ff5e

feat: expose chunk dissasemble d9b2f22710

feat: f64 to string dc19d70838

feat: upgrade to zig 0.16.0 4ef408c763

Merge pull request 'feat: upgrade to zig 0.16.0' (#4 ) from feat/operators into develop 3c1ac7747d

Reviewed-on: #4

feat: begin work on functions 771f574188

feat: helper for object downcasting 496dc63ced

feat: pub object & object function refs cc60bba012

chore: build error 0c471166b4

chore: const str 8270567a60

feat: ideas on new vm design 169697551c

chore: rename some vars 9d578a0000

feat(circles): add rip/rsp/rbp & stack to the VM 5d4283893d

feat: rework to have a list of 'callframes' instead of a chunk 5445bac115

fix: use frame to access chunk for now d61ecffdf3

feat: op_call opcode - incomplete 44e3747e1c

feat: op_call minimal impl efcf5d9895

Pushes needed RIP/RSP to the stack, updates RIP/RSP/RBP for the new
frame

feat: improve op_call & implement op_ret f80981db75

On op_call setup better registers, fix some pointer bugs. Implement
op_ret to handle the entrypoint or additional frames.

Merge pull request 'Minimal implementation of functions' (#5 ) from feat/functions into develop 30fda12caf

Reviewed-on: #5

feat: fix tests 02e4c7b046

Fixes tests broken & left unfixed after the 0.15.1 writergate
scandal.

test: improve tests & docs for object.zig 257d1de9bd

feat: add option to build only in library mode e07b431ef2

fix: enable chunk methods on tests too 9279bb0ef4

fix: leak on VM new frames field 3d0f9afdcd

test: function call path fc3e149518

This commit creates a test for calling a "function" (create a chunk,
store it, load it, .op_call & .op_ret).

However, it revealed a limitation: for a ObjectFunction to receive a
Chunk, this last one **must** be created with the same allocator, for
the ObjectFunction will `destroy` it. This means one cant use a local
variable, or a single chunk from an array, or other allocation patterns
other than `destroy`.

This could be fixed by always requiring the caller to pass a chunk, and
leaving cleanup to them. Will investigate further.

chore: small changes 5d901bd652

Merge pull request 'feat/tiger' (#6 ) from feat/tiger into develop 9c285e1204

Reviewed-on: #6

wip: rework chunk, move chunk allocations (string, object) to VM 11e23823af

rework: new chunk format & serialization a6e420bbb6

- parse & fmt bytecode header
- parse values section len
- parse blob section
- parse constants section & log
- parse functions section

all: wire new vmfile into main flow bcb5b9f1a6

- read vmfile from stdin
- parse vmfile
- create VM
- run VM with code from vmfile
- mark regression on **all** opcodes

vm: reenable opcodes e680fa9fc8

- Resolve identities through vm manager
- restore load opcodes
- restore opcodes that deal w registers
- dissasemble opcodes on deserialization debug

all: merge aa9d40121d

serialization(AI): improve code/tests/deserialization 73e9521965

- use proper name for disassemble & related 
- serialize to disk
- finish opcode dissasembly
- move human readable printing of vmfile to its own file

all: remove old Chunk 361f9c6da3

cli(AI): implement a CLI interface for main 803a40f5e1

serialization: add a module name to a vmfile 633aba8f47

serialization(AI): fix tests to include new module name in vmfile f3e0eea096

fernando merged commit ad03984586 into main

2026-06-23 21:35:54 +00:00

fernando referenced this pull request from a commit

2026-06-23 21:35:55 +00:00

Merge pull request 'rework vm file format' (#7) from develop into main

No reviewers

No labels

No milestone

No project

No assignees

1 participant

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

fernando/naravm!7

No description provided.

Rows
Columns