V0.2.3/spatial streaming - #47
Merged
Merged
Conversation
…ntal I/O.
**Behavior**
`source: :file` (or `:auto` path detection): read 16-byte EXCP / 18-byte GSPL header, then one fixed-size record via `IO.binread/2`
Peak memory ≈ header + one record (file handle stays open for the stream lifetime)
In-memory binaries still materialize through `decode/2` (mmap left for a future Rust backend)
**Wiring**
`ExCodecs.Spatial.Stream` forwards `EXCP/GSPL` sources to the codecs — no more full File.read for those formats
Docs updated in `spatial.ex`, `docs/spatial_formats.md`, the spatial guide, and `CHANGELOG` Unreleased
**Usage**
```elixir
ExCodecs.Spatial.stream_decode(path, format: :spatial_binary, source: :file)
|> Stream.take(100)
|> Enum.to_list()
ExCodecs.Spatial.stream_decode(path, format: :gsplat, source: :file)
|> Enum.each(&process/1)
```
Truncation / missing files yield a single `{:error, %ExCodecs.Error{}}` element.
closed #38 - PLY stream_decode (source: :file) — scans until end_header, then yields one vertex at a time (binary: fixed stride via IO.binread; ASCII: line reads). Supports :as / Gaussian auto-detect.
closed #39 - stream_encode_to_file/3 — EXCP and GSPL write a placeholder header, stream records, seek back to patch count. Requires explicit :schema. Spatial.Stream.encode_to_file/3 uses this when schema: + format: :spatial_binary / :gsplat.
closed #40
…plus enumerable_points/1 / enumerable_gaussians/1. Credo is clean and the stream tests pass.
Version
mix.exs → 0.2.3
native/ex_codecs_native/Cargo.toml → 0.2.3
Changelog
Moved Unreleased items into [0.2.3] - 2026-07-18 (streaming I/O, schema encode-to-file, Rust accel, Mix cli/0)
Docs
README install pin, streaming limitations, native layer note
docs/spatial_formats.md, docs/architecture.md, guides/understanding_spatial_codecs.md
All livebooks → {:ex_codecs, "~> 0.2.3"}; spatial livebook streaming section updated
Fix NIF availability registration order, drop stale docs from the Hex package, correct streaming/work_factor/from_nif docs, repair livebook demos, and make Accel tests skip cleanly when the spatial NIF is absent. Co-authored-by: Cursor <cursoragent@cursor.com>
…xed in this final pass:
Code fixes
- A-5: Fixed doc example to use MyApp.CodecRegistry instead of the global ExCodecs.CodecRegistry (avoids name collision)
- CR-1: Added Code.ensure_loaded?/1 guard in build_codec_info/5 before probing __codec_info__/0
- CR-4: Documented the Agent's purpose (owns ETS table, state never read)
- C-3: Documented that validates?/1 checks interface conformance, not data validation
- C-4: Annotated Zstd typedoc examples with NIF precondition
- Z-1: Documented the Elixir/Rust level-clamp divergence in Zstd encode docs
- Z-2: zstd_version/0 now reads Native.codec_versions() with a static fallback
- ER-1: Unknown NIF error atoms now map to :invalid_data with the raw atom in details:
- ER-2: Widened matches?/2 spec to term() first argument
- NF-2: Removed hardcoded 268435456 from the :output_limit_exceeded message
- NF-3: Corrected safe_call/2 spec to {:ok, binary()}
- N-2: Commented the nif_loaded?/0 rescue clauses with concrete scenarios
- N-3: Documented the base_url ↔ @source_url coupling in native.ex
- SP-5: resolve_as/2 now returns {:error, :invalid_options} instead of raising CaseClauseError
- SP-6: Replaced all em-dashes and non-breaking hyphens with plain hyphens in spatial docs
- SP-9: Marked Transform as "Reserved for future use" in its moduledoc
- SP-10: Documented the atom-key restriction for literal %Point{} construction
- SP-11: Added numeric guards to Bounds.new/2
- SP-12: Added sh shape validation in Gaussian.new/2
- SP-15: Restricted safe/1 rescue to expected exception classes with warning logging
- SP-16: Removed true as an accepted endianness atom
- SP-19: EXCP decode now rejects unknown flag bits with :invalid_data
- SP-20: Collapsed 3-pass Enum.any? + length/1 into a single Enum.reduce
- SP-23: Restricted PLY encode rescue to ArithmeticError/FunctionClauseError/ArgumentError
- SP-24: Dropped redundant metadata: attrs in PLY Gaussian decode
- SP-25: Added validate_comments/1 rejecting newlis and end_header in PLY comments
- SP-27: Documented numeric-only attribute restriction in Point typedoc
- R-5/R-6: Documented partial-record caller obligation and color/SH coercion in Accel moduledoc
- R-8: Documented the N×256 MiB amplification risk in Compression.decompress/3
Documentation/test fixes
- D-1: Verified repo slug thanos/codecs is consistent with git remote
- D-2: Removed duplicate checksum reminder from 0.2.0 CHANGELOG
- D-5: Added Keep a Changelog link references at bottom of CHANGELOG
- D-10: Fixed codec chaining recommendation (spatial→zstd, not zstd→blosc2)
- T-13/T-17/T-18: Verified already fixed (exact pins, split tests, tmp cleanup)
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
stream_decodefrom files (bounded memory), plus schema-drivenstream_encode_to_file