Repository navigation
Upgrade vendored OCaml to 5.5.0 and merlin to 5.8.1-505 - #44
Merged
Merged
Conversation
- bump ocaml submodule 5.1.0 -> 5.5.0 - re-vendor parsing sources via copy_files promotion - port JSX grammar/lexer patches onto the 5.5.0 parser.mly/lexer.mll - target Astlib.Ast_505 (ppxlib >= 0.38 provides it) - adapt jsx_helper/mlx_shim to the new Longident.t with located components, Format_doc-based Location.errorf/msg, Pprintast.Doc Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- bump merlin submodule v4.14-501 -> v5.8.1-505 - re-vendor merlin sources via copy_files promotion; drop files removed upstream (utils/misc, utils/sys, typing/rec_check) - port JSX patches onto 505 parser_raw.mly/lexer_raw.mll - regenerate parser_raw/parser_recover/parser_printer - bump menhir dev pin 20201216 -> 20230608 (version merlin 5.8.1 vendors as menhirLib); update root menhirLib copy to match - target Astlib.Ast_505; reorder opens so merlin Config wins over astlib's in mlx_ocaml_parsing - sync mconfig_dot stub with 505 config shape; adapt frontend to the new Parsetree.constant record and Format_doc-based error reports - accept 5.5-style unquoted code spans in error-message test output Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- ppxlib lower bound 0.34 -> 0.38 (first release with Ast_505) - constrain merlin-lib >= 5.8.1 - CI: add 5.5 to the compiler matrix; bump menhir install to 20230608 - Makefile: create-switch now uses 5.5.0 Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
On 5.5 the solver can only pick 5.8.1-505 anyway; on older compilers the build fails in the vendored sources regardless (they use the 5.5 stdlib), so the constraint only moved the failure, and it offers no protection against future merlin-lib releases either. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The vendored 5.5 sources referenced host compiler-libs APIs that only
exist on a 5.5 host (where astlib aliases to compiler-libs). Shim them
locally so the package builds on 4.14-5.4 hosts too:
- vendor the compiler's format_doc.{ml,mli} (with a local
utf_8_scalar_width fallback, stdlib >= 5.4) into mlx_shim
- Misc.Style, Misc.Utf8_lexeme, Pprintast.Doc.tyvar,
Clflags.{keyword_edition,parse_keyword_edition} shims
- no-op Builtin_attributes (w53 bookkeeping is the consuming
compiler's job, not the preprocessor's)
- shadow Ocaml_common.Longident with the astlib one: menhir --infer
bakes host-canonical type paths into the generated parser on a 5.5
host, and this makes that spelling portable
Verified: dev+release build and tests on 5.5; release build on 5.4 and
4.14 host switches, with correct mlx-pp output from the 4.14 binary.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- fix the reader's Conv bridge, still on the Ast_501/OCaml_501 path from before the upgrade: with 505-shaped memory that Obj.magic chain was a latent runtime bug even on 5.5 (the cram tests bypass Conv); now converts 505 -> host version via ppxlib migrations - shim stdlib APIs the vendored merlin 505 sources use, as local modules in mlx_utils reached through the existing -open chain: Seq.delay (5.5), Pair (5.4), String.edit_distance (5.4), Format.utf_8_scalar_width (5.4), Map.S of_list/to_list/add_to_list (5.1), Sys.is_regular_file (5.1), Digest.BLAKE128 (5.2, MD5-backed; only the type is used and a reader never populates consistbl) - stub Merlin_index_format (absent from merlin-lib 4.x): the kernel only flushes its cache, which a reader never populates Verified: 5.5 dev+release+tests; release builds on 5.4 (merlin-lib 5.7.1-504) and 4.14 (merlin-lib 4.19-414) host switches; the merlin reader driver produces correct JSX output on both. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The cram MERLIN sections print the vendored AST directly, so Mlx_conv — the Obj.magic + ppxlib migration bridge the reader actually uses to hand its parsetree to the host merlin — had no coverage; the stale Ast_501 bridge fixed earlier segfaulted yet all tests were green. - extract Conv from the frontend into an mlx_conv library - mlx_merlin.exe -conv (and -intf) runs the parse through Mlx_conv and prints via the HOST merlin-lib Pprintast, so a shape mismatch segfaults or garbles the output (verified: swapping the bridge back to OCaml_501 makes these tests crash with SIGSEGV) - cram cases cover constants (the representation that changed across AST versions: int, string, float, char), nested JSX, punned and optional props, patterns, and signature conversion Also verified by hand on 5.4 and 4.14 host switches, where the bridge performs real 505->504 / 505->414 migrations. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
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.
Upgrades the vendored submodules so the mlx dialect gains OCaml 5.5 syntax (effect handlers, etc.):
Astlib.Ast_505)Verified locally on a 5.5.0 switch: dev + release builds green, full cram suite passes (batch and merlin paths).
Draft to collect CI matrix signal on 4.14 / 5.4 / 5.5 — the older legs are expected to fail; their failures scope the backward-compat work.
🤖 Generated with Claude Code