glob and fnmatch

[adh]

No description provided.

[Copilot]

Pull request overview

This PR adds POSIX-style filename pattern matching and globbing to ListTalk, exposing them both as ListTalk:OS module primitives and as convenience methods on String, with accompanying evaluation tests.

Changes:

• Add ListTalk:OS:fnmatch? (keyword-flagged) and ListTalk:OS:glob primitives backed by libc fnmatch(3) / glob(3).
• Add String>>fnMatch?: and String>>fnMatchPathname?: methods for common matching modes.
• Add evaluation tests covering default vs pathname matching and basic glob behavior.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 3 comments.

File	Description
tests/eval-objects.lt	Adds tests for new String pattern-matching methods.
tests/eval-modules-os.lt	Adds tests for ListTalk:OS:fnmatch? and ListTalk:OS:glob.
src/modules/os.c	Implements fnmatch? and glob primitives and keyword-to-flag mapping.
src/classes/String.c	Implements two new String methods backed by fnmatch().

Suppressed comments (1)

src/classes/String.c:1033

• String>>fnMatchPathname?: also treats any nonzero fnmatch() result as a non-match. Consider distinguishing FNM_NOMATCH from other error returns and raising for the latter so invalid patterns/flags don't get swallowed.

    return fnmatch(
        LT_String_value_cstr(pattern),
        LT_String_value_cstr(self),
        FNM_PATHNAME
    ) == 0 ? LT_TRUE : LT_FALSE;

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated no new comments.

Suppressed comments (3)

src/modules/os.c:954

• primitive_os_glob currently returns an empty list for all non-zero glob() return codes, which masks real failures (e.g. GLOB_ABORTED/GLOB_NOSPACE) and makes flags like :err ineffective. Only GLOB_NOMATCH should be silently mapped to an empty list; other errors should raise.

    result = glob(LT_String_value_cstr(pattern), flags, NULL, &matches);
    builder = LT_ListBuilder_new();
    if (result == 0){
        for (index = 0; index < matches.gl_pathc; index++){
            LT_ListBuilder_append(

src/modules/os.c:9

• _GNU_SOURCE is currently defined unconditionally for this translation unit. Elsewhere (e.g. src/classes/Thread.c:6-15) feature-test macros are gated per-platform; defining _GNU_SOURCE on non-glibc targets can inadvertently change header visibility/ABI and may trigger warnings.

#ifndef _GNU_SOURCE
#define _GNU_SOURCE
#endif

src/modules/os.c:931

• The docstring says "glob errors produce an empty list", but the tests/expected behavior only mention ignoring the no-match case. If non-NOMATCH errors are intended to be surfaced (e.g. :err, OOM, permission errors), the docstring should reflect that more precisely.

This issue also appears on line 950 of the same file.

    "Return paths matching a POSIX glob pattern; glob errors produce an empty list."