Add C workflow docs and streamlined run builtin - #105
Conversation
Wire the run builtin into the shell, add a staged data-oriented compile/build/run pipeline for C file and directory targets, and enforce clang-family toolchain selection for BSD/Apache-2.0 alignment. Update run tests for validation, streamlined argv passthrough, and directory project builds, and align CLI spec docs with the implemented run syntax/options. Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Organization UI Review profile: ASSERTIVE Plan: Pro Plus Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
📝 WalkthroughWalkthroughThe change introduces a cached ChangesC workflow and shell modernization
Estimated code review effort: 5 (Critical) | ~120 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 16
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
sh.init.c (1)
43-153: 🎯 Functional Correctness | 🟠 Major | ⚡ Quick winRestore the unrelated platform builtin registrations.
This change removes guarded builtins such as
att,bye,getspath,getxvers,migrate, anduniverse. Addingrunshould not silently remove supported commands on their respective platforms.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@sh.init.c` around lines 43 - 153, Restore the platform-guarded builtin registrations removed from the builtin table, including att, bye, getspath, getxvers, migrate, and universe, using their original conditional compilation guards and handlers. Keep the new run entry and existing alphabetical ordering intact, ensuring each platform retains its supported commands.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@docs/c-commands-spec.md`:
- Around line 1-3: Update the documentation to reflect delivered P1 support: in
docs/c-commands-spec.md (lines 1-3) describe run as implemented while keeping
compile/build planned; in docs/c-workflow-analysis.md (lines 1-3) distinguish
completed P1 behavior from future phases; in docs/compiler/ARCHITECTURE.md
(lines 1-4) update status and remaining integration work; in
docs/compiler/CACHE-DESIGN.md (lines 1-4) distinguish the implemented cache
layout from future index/metadata features; in docs/compiler/CLI-SPEC.md (lines
1-17) mark compile/build syntax as planned and document only supported run
options; in docs/compiler/DATA-MODEL.md (lines 1-4) identify the model as a
future target beyond the initial implementation; in
docs/compiler/DEPENDENCY-DISCOVERY.md (lines 1-4) describe dependency scanning
as future work; and in docs/compiler/INDEX.md (lines 1-5) remove the claim that
no runtime is implemented.
In `@docs/compiler/CACHE-DESIGN.md`:
- Around line 55-77: Update the cache-key design in the documented key
construction and normalization section to preserve order and multiplicity:
replace XOR of object keys with a length-delimited ordered sequence hash, and
hash the effective linker/compiler argv in its original order. Remove sorting of
order-sensitive linker flags and canonicalized -I paths, while retaining only
normalization that does not change argument order or meaning.
In `@docs/compiler/DECISIONS.md`:
- Around line 223-227: Update the ADR-006 execution-model documentation around
the `run` behavior to match the implementation: `cw_run_execute_stage()` runs
the stage in a child, returns its status to `dorun()`, and `dorun()` updates
shell status and performs cleanup. Remove or correct the recommendation to use
direct `execvp` so the documented model preserves fork-and-wait behavior and
returns control to the pipeline.
In `@docs/compiler/DEPENDENCY-DISCOVERY.md`:
- Around line 145-158: Revise the “mtime-gated hashing” section to remove the
claim that matching mtime and size safely validates content_hash. Require
hashing of cache-key inputs for correctness, or explicitly document the
optimization as best-effort and add a verification path before accepting the
cached hash.
In `@docs/compiler/MASTER-PLAN.md`:
- Around line 3-4: Synchronize the compiler planning documents with the
implemented PR: in docs/compiler/MASTER-PLAN.md at lines 3-4, replace the
no-runtime status and update the corresponding P1 checklist; in
docs/compiler/MILESTONES.md at lines 3-4, replace the planning-only status and
accurately mark the shipped M1 work.
In `@docs/compiler/PIPELINES.md`:
- Around line 291-303: Update the Stage 4 “Execute” section to document the
shipped fork+exec+waitpid behavior in cw_execute_binary/cw_run_execute_stage
rather than direct execvp replacement. Describe the forked child executing the
binary, the parent waiting and propagating its status, and retain the
exec-failure exit code 127; remove claims that run replaces the mcsh process
directly or requires avoiding an intermediate shell process.
In `@ISSUES.md`:
- Around line 9-21: Refresh the compiler planning documentation to match the
shipped dorun() implementation: in ISSUES.md lines 9-21, move CW-001 from Open
to Resolved and note its directory-target support; in PLAN.md lines 48-57, mark
M1/run as implemented and note directory/project support moved forward from M3;
in docs/compiler/PIPELINES.md lines 1-6, remove or revise the
planning-only/no-implementation status for the run pipeline; and in
docs/compiler/TODO.md lines 28-83, check off the completed P1 items for
sh.cworkflow.c, dorun(), sh.decls.h/sh.init.c wiring, and the run tests.
In `@sh.cworkflow.c`:
- Around line 959-978: Update cw_execute_binary and the cw_run_execute_stage
handling to distinguish an execv failure from a user program that exits 127,
mirroring the existing compile/build-stage status == 127 heuristic. Emit the
required “run: exec failed: <strerror>” diagnostic using the captured execution
error, preserve exit code 127, and keep the existing fork-failure handling
unchanged.
- Around line 807-826: Document the known limitation that project_hash, used by
cw_compute_object_path, is incorporated into every object cache key for
directory targets, causing all objects to rebuild when any discovered source or
header changes. Add this limitation to the appropriate RISKS.md or PLAN.md
Phase-3/known-limitations section, noting that precise incremental caching
awaits include-dependency discovery; do not change the hashing behavior.
- Around line 828-840: Update cw_hash_toolchain and its callers so repeated run
invocations do not SHA-256-hash the entire compiler binary; cache the toolchain
identity keyed by resolved compiler path, modification time, and size, and reuse
it when those attributes are unchanged. Recompute and refresh the cached hash
only when the key changes, while preserving the existing fallback behavior for
hashing the path when file hashing fails.
In `@tcsh.man.in`:
- Around line 8909-8913: Correct the category-count statement in the
compile-time default-colors documentation near the listed token categories:
either update the enumeration to include all 12 categories or remove the
unsupported numeric count, while keeping the existing category descriptions
accurate.
- Around line 8483-8496: Update the %G operation-state documentation to match
the detection logic in git_get_info: remove REBASE_HEAD and
rebase-merge/head-name, and document rebase-apply/rebasing, rebase-apply (AM),
REVERT_HEAD, and BISECT_LOG with their corresponding states.
In `@tests/t023_filetest_ops.sh`:
- Around line 4-12: The test fixture around the C-shell -z check currently uses
an empty tmpfile while expecting ! -z to print z_ok. Update the fixture or
expectation in the test command so it matches correct -z semantics, preserving
the existing checks for -d, -f, and -e.
In `@tests/t100_run_validation.sh`:
- Around line 37-48: Update the non-clang compiler enforcement test around the
gcc availability check to run unconditionally using a guaranteed non-clang
executable such as /bin/sh. Remove the environment-dependent skip, while
preserving the non-zero status assertion and “not a clang-family compiler”
message validation; if no suitable executable can be selected, fail the test
explicitly.
In `@tests/t101_run_basic.sh`:
- Around line 66-71: Strengthen the second invocation in the cache reuse test
around the mcsh command so it runs with verbose output and asserts that no
compile or link actions are emitted, while preserving the existing status and
“hello” output checks. Verify cached artifacts are reused rather than merely
confirming repeated execution correctness.
- Around line 80-85: Extend the run argument-passthrough test around the
existing status/output assertion to also invoke run with the `--` separator
before `one two`. Assert that this dedicated parsing path returns success and
produces exactly `one two`, using the same failure reporting and exit behavior
as the existing direct-forwarding check.
---
Outside diff comments:
In `@sh.init.c`:
- Around line 43-153: Restore the platform-guarded builtin registrations removed
from the builtin table, including att, bye, getspath, getxvers, migrate, and
universe, using their original conditional compilation guards and handlers. Keep
the new run entry and existing alphabetical ordering intact, ensuring each
platform retains its supported commands.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: ASSERTIVE
Plan: Pro Plus
Run ID: 64a83388-2554-4b51-91d0-ef216852d958
📒 Files selected for processing (47)
ISSUES.mdMakefile.inPLAN.mdREADME.mdSYNTAX_AND_SCRIPTING.mddocs/c-commands-spec.mddocs/c-workflow-analysis.mddocs/compiler/ARCHITECTURE.mddocs/compiler/CACHE-DESIGN.mddocs/compiler/CLI-SPEC.mddocs/compiler/DATA-MODEL.mddocs/compiler/DECISIONS.mddocs/compiler/DEPENDENCY-DISCOVERY.mddocs/compiler/INDEX.mddocs/compiler/MASTER-PLAN.mddocs/compiler/MILESTONES.mddocs/compiler/PIPELINES.mddocs/compiler/RISKS.mddocs/compiler/TEST-PLAN.mddocs/compiler/TODO.mdfix_truncation.patchsh.csh.cworkflow.csh.decls.hsh.dir.csh.func.csh.init.csh.sem.cstrncpy_analysis.mdtc.prompt.ctc.prompt.c.origtcsh.man.intest2test3tests/t005_cd_stack.shtests/t015_dotmcshrc_ls_colors.shtests/t017_cd_minus_n.shtests/t018_syntax_highlight.shtests/t019_git_prompt.shtests/t020_predict.shtests/t021_onintr.shtests/t022_function_args.shtests/t023_filetest_ops.shtests/t024_variable_modifiers.shtests/t025_switch.shtests/t100_run_validation.shtests/t101_run_basic.sh
💤 Files with no reviewable changes (3)
- strncpy_analysis.md
- tc.prompt.c.orig
- fix_truncation.patch
| static int | ||
| cw_hash_file_set(const cw_str_list_t *files, uint8_t out[32]) | ||
| { | ||
| cw_sha256_ctx_t ctx; | ||
| uint8_t file_hash[32]; | ||
| size_t i; | ||
| char nul = '\0'; | ||
|
|
||
| cw_sha256_init(&ctx); | ||
| for (i = 0; i < files->len; i++) { | ||
| if (cw_sha256_file(files->items[i], file_hash) != 0) | ||
| return -1; | ||
| cw_sha256_update(&ctx, (const uint8_t *)files->items[i], | ||
| strlen(files->items[i])); | ||
| cw_sha256_update(&ctx, (const uint8_t *)&nul, 1); | ||
| cw_sha256_update(&ctx, file_hash, sizeof(file_hash)); | ||
| } | ||
| cw_sha256_final(&ctx, out); | ||
| return 0; | ||
| } |
There was a problem hiding this comment.
🚀 Performance & Scalability | 🟠 Major | 🏗️ Heavy lift
Whole-project hash baked into every object's cache key causes full rebuilds on any single-file change for directory targets.
project_hash (an aggregate SHA-256 over every discovered .c/.h file — line 1135) is mixed into every individual object's cache key in cw_compute_object_path (lines 858-859). Since any file edit anywhere in the tree changes project_hash, it invalidates every object's cache key, not just the unit(s) that actually depend on the changed file — forcing a full recompile of the whole project on any single-file edit for run <directory> targets. This is safe (no wrong results) but defeats the incremental-caching value the project's own docs describe (e.g. TEST-PLAN.md I111 "Only the changed file recompiles").
This is presumably an accepted stopgap in the absence of real #include dependency discovery (CW-002/M2), but it's worth calling out explicitly as a known limitation (e.g., in RISKS.md or PLAN.md's Phase-3/known-limitations list) so it isn't mistaken for working incremental caching once directory run support is exercised on non-trivial projects.
Also applies to: 842-865, 1135-1135
🧰 Tools
🪛 Clang (14.0.6)
[warning] 812-812: variable 'i' is not initialized
(cppcoreguidelines-init-variables)
[warning] 812-812: variable name 'i' is too short, expected at least 3 characters
(readability-identifier-length)
[warning] 817-817: statement should be inside braces
(readability-braces-around-statements)
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@sh.cworkflow.c` around lines 807 - 826, Document the known limitation that
project_hash, used by cw_compute_object_path, is incorporated into every object
cache key for directory targets, causing all objects to rebuild when any
discovered source or header changes. Add this limitation to the appropriate
RISKS.md or PLAN.md Phase-3/known-limitations section, noting that precise
incremental caching awaits include-dependency discovery; do not change the
hashing behavior.
| static int | ||
| cw_hash_toolchain(const char *cc_path, uint8_t out[32]) | ||
| { | ||
| cw_sha256_ctx_t ctx; | ||
|
|
||
| if (cw_sha256_file(cc_path, out) == 0) | ||
| return 0; | ||
|
|
||
| cw_sha256_init(&ctx); | ||
| cw_sha256_update(&ctx, (const uint8_t *)cc_path, strlen(cc_path)); | ||
| cw_sha256_final(&ctx, out); | ||
| return 0; | ||
| } |
There was a problem hiding this comment.
🚀 Performance & Scalability | 🟠 Major | 🏗️ Heavy lift
Hashing the entire compiler binary on every run invocation conflicts with the documented <50ms warm-cache-hit target.
cw_hash_toolchain() SHA-256-hashes the whole resolved compiler binary (clang/cc, often 50-150+ MB) unconditionally in cw_run_prepare_state, on every single run call — including a fully cached, nothing-to-compile invocation. TEST-PLAN.md §6.1 and PIPELINES.md Stage 2 both call out a <50ms warm-cache-hit budget; hashing a large binary end-to-end will typically blow well past that on every invocation, defeating the "fast cache check" design goal.
Consider caching the toolchain identity hash itself (e.g., keyed by resolved path + mtime + size in a small sidecar/index, recomputed only when those change) instead of re-hashing the full binary content every time.
Also applies to: 1110-1114
🧰 Tools
🪛 Clang (14.0.6)
[warning] 833-833: statement should be inside braces
(readability-braces-around-statements)
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@sh.cworkflow.c` around lines 828 - 840, Update cw_hash_toolchain and its
callers so repeated run invocations do not SHA-256-hash the entire compiler
binary; cache the toolchain identity keyed by resolved compiler path,
modification time, and size, and reuse it when those attributes are unchanged.
Recompute and refresh the cached hash only when the key changes, while
preserving the existing fallback behavior for hashing the path when file hashing
fails.
| if command -v gcc >/dev/null 2>&1; then | ||
| out=$("$MCSH" -f -c "set mcsh_cc = gcc; run $tmpdir/hello.c" 2>&1) | ||
| status=$? | ||
| if [ $status -eq 0 ]; then | ||
| printf 'expected non-zero status for non-clang mcsh_cc, got 0\n' | ||
| exit 1 | ||
| fi | ||
| printf '%s\n' "$out" | grep -F "not a clang-family compiler" >/dev/null 2>&1 || { | ||
| printf "expected clang-family enforcement message, got: %s\n" "$out" | ||
| exit 1 | ||
| } | ||
| fi |
There was a problem hiding this comment.
📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win
Make the non-clang override test unconditional.
Clang-only environments skip the sole $mcsh_cc enforcement test. Use a guaranteed non-clang executable such as /bin/sh, or explicitly fail if no suitable candidate exists.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@tests/t100_run_validation.sh` around lines 37 - 48, Update the non-clang
compiler enforcement test around the gcc availability check to run
unconditionally using a guaranteed non-clang executable such as /bin/sh. Remove
the environment-dependent skip, while preserving the non-zero status assertion
and “not a clang-family compiler” message validation; if no suitable executable
can be selected, fail the test explicitly.
| out=$("$MCSH" -f -c "set mcsh_cache_dir = $tmpdir/cache; run $tmpdir/hello.c" 2>&1) | ||
| status=$? | ||
| if [ $status -ne 0 ] || [ "$out" != "hello" ]; then | ||
| printf "expected cached hello/0, got status=%d output=%s\n" "$status" "$out" | ||
| exit 1 | ||
| fi |
There was a problem hiding this comment.
📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win
Assert cache reuse rather than repeated correctness.
The second invocation only verifies that the program still prints hello; recompiling and relinking every time would also pass. Assert that the second -v run emits no compile/link action, or otherwise verify cached artifacts are reused.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@tests/t101_run_basic.sh` around lines 66 - 71, Strengthen the second
invocation in the cache reuse test around the mcsh command so it runs with
verbose output and asserts that no compile or link actions are emitted, while
preserving the existing status and “hello” output checks. Verify cached
artifacts are reused rather than merely confirming repeated execution
correctness.
| out=$("$MCSH" -f -c "set mcsh_cache_dir = $tmpdir/cache; run $tmpdir/args.c one two" 2>&1) | ||
| status=$? | ||
| if [ $status -ne 0 ] || [ "$out" != "one two" ]; then | ||
| printf "expected streamlined arg passthrough, got status=%d output=%s\n" "$status" "$out" | ||
| exit 1 | ||
| fi |
There was a problem hiding this comment.
🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win
Cover the -- separator path.
This checks direct argument forwarding, but not the dedicated -- parsing branch promised by the run interface. Add an equivalent run … -- one two assertion.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@tests/t101_run_basic.sh` around lines 80 - 85, Extend the run
argument-passthrough test around the existing status/output assertion to also
invoke run with the `--` separator before `one two`. Assert that this dedicated
parsing path returns success and produces exactly `one two`, using the same
failure reporting and exit behavior as the existing direct-forwarding check.
There was a problem hiding this comment.
Caution
Inline review comments failed to post. This is likely due to GitHub's internal server error or limits when posting large numbers of comments. If you are seeing this consistently it is likely a permissions issue. Please check "Moderation" -> "Code review limits" under your organization settings.
Actionable comments posted: 16
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
sh.init.c (1)
43-153: 🎯 Functional Correctness | 🟠 Major | ⚡ Quick winRestore the unrelated platform builtin registrations.
This change removes guarded builtins such as
att,bye,getspath,getxvers,migrate, anduniverse. Addingrunshould not silently remove supported commands on their respective platforms.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@sh.init.c` around lines 43 - 153, Restore the platform-guarded builtin registrations removed from the builtin table, including att, bye, getspath, getxvers, migrate, and universe, using their original conditional compilation guards and handlers. Keep the new run entry and existing alphabetical ordering intact, ensuring each platform retains its supported commands.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@docs/c-commands-spec.md`:
- Around line 1-3: Update the documentation to reflect delivered P1 support: in
docs/c-commands-spec.md (lines 1-3) describe run as implemented while keeping
compile/build planned; in docs/c-workflow-analysis.md (lines 1-3) distinguish
completed P1 behavior from future phases; in docs/compiler/ARCHITECTURE.md
(lines 1-4) update status and remaining integration work; in
docs/compiler/CACHE-DESIGN.md (lines 1-4) distinguish the implemented cache
layout from future index/metadata features; in docs/compiler/CLI-SPEC.md (lines
1-17) mark compile/build syntax as planned and document only supported run
options; in docs/compiler/DATA-MODEL.md (lines 1-4) identify the model as a
future target beyond the initial implementation; in
docs/compiler/DEPENDENCY-DISCOVERY.md (lines 1-4) describe dependency scanning
as future work; and in docs/compiler/INDEX.md (lines 1-5) remove the claim that
no runtime is implemented.
In `@docs/compiler/CACHE-DESIGN.md`:
- Around line 55-77: Update the cache-key design in the documented key
construction and normalization section to preserve order and multiplicity:
replace XOR of object keys with a length-delimited ordered sequence hash, and
hash the effective linker/compiler argv in its original order. Remove sorting of
order-sensitive linker flags and canonicalized -I paths, while retaining only
normalization that does not change argument order or meaning.
In `@docs/compiler/DECISIONS.md`:
- Around line 223-227: Update the ADR-006 execution-model documentation around
the `run` behavior to match the implementation: `cw_run_execute_stage()` runs
the stage in a child, returns its status to `dorun()`, and `dorun()` updates
shell status and performs cleanup. Remove or correct the recommendation to use
direct `execvp` so the documented model preserves fork-and-wait behavior and
returns control to the pipeline.
In `@docs/compiler/DEPENDENCY-DISCOVERY.md`:
- Around line 145-158: Revise the “mtime-gated hashing” section to remove the
claim that matching mtime and size safely validates content_hash. Require
hashing of cache-key inputs for correctness, or explicitly document the
optimization as best-effort and add a verification path before accepting the
cached hash.
In `@docs/compiler/MASTER-PLAN.md`:
- Around line 3-4: Synchronize the compiler planning documents with the
implemented PR: in docs/compiler/MASTER-PLAN.md at lines 3-4, replace the
no-runtime status and update the corresponding P1 checklist; in
docs/compiler/MILESTONES.md at lines 3-4, replace the planning-only status and
accurately mark the shipped M1 work.
In `@docs/compiler/PIPELINES.md`:
- Around line 291-303: Update the Stage 4 “Execute” section to document the
shipped fork+exec+waitpid behavior in cw_execute_binary/cw_run_execute_stage
rather than direct execvp replacement. Describe the forked child executing the
binary, the parent waiting and propagating its status, and retain the
exec-failure exit code 127; remove claims that run replaces the mcsh process
directly or requires avoiding an intermediate shell process.
In `@ISSUES.md`:
- Around line 9-21: Refresh the compiler planning documentation to match the
shipped dorun() implementation: in ISSUES.md lines 9-21, move CW-001 from Open
to Resolved and note its directory-target support; in PLAN.md lines 48-57, mark
M1/run as implemented and note directory/project support moved forward from M3;
in docs/compiler/PIPELINES.md lines 1-6, remove or revise the
planning-only/no-implementation status for the run pipeline; and in
docs/compiler/TODO.md lines 28-83, check off the completed P1 items for
sh.cworkflow.c, dorun(), sh.decls.h/sh.init.c wiring, and the run tests.
In `@sh.cworkflow.c`:
- Around line 959-978: Update cw_execute_binary and the cw_run_execute_stage
handling to distinguish an execv failure from a user program that exits 127,
mirroring the existing compile/build-stage status == 127 heuristic. Emit the
required “run: exec failed: <strerror>” diagnostic using the captured execution
error, preserve exit code 127, and keep the existing fork-failure handling
unchanged.
- Around line 807-826: Document the known limitation that project_hash, used by
cw_compute_object_path, is incorporated into every object cache key for
directory targets, causing all objects to rebuild when any discovered source or
header changes. Add this limitation to the appropriate RISKS.md or PLAN.md
Phase-3/known-limitations section, noting that precise incremental caching
awaits include-dependency discovery; do not change the hashing behavior.
- Around line 828-840: Update cw_hash_toolchain and its callers so repeated run
invocations do not SHA-256-hash the entire compiler binary; cache the toolchain
identity keyed by resolved compiler path, modification time, and size, and reuse
it when those attributes are unchanged. Recompute and refresh the cached hash
only when the key changes, while preserving the existing fallback behavior for
hashing the path when file hashing fails.
In `@tcsh.man.in`:
- Around line 8909-8913: Correct the category-count statement in the
compile-time default-colors documentation near the listed token categories:
either update the enumeration to include all 12 categories or remove the
unsupported numeric count, while keeping the existing category descriptions
accurate.
- Around line 8483-8496: Update the %G operation-state documentation to match
the detection logic in git_get_info: remove REBASE_HEAD and
rebase-merge/head-name, and document rebase-apply/rebasing, rebase-apply (AM),
REVERT_HEAD, and BISECT_LOG with their corresponding states.
In `@tests/t023_filetest_ops.sh`:
- Around line 4-12: The test fixture around the C-shell -z check currently uses
an empty tmpfile while expecting ! -z to print z_ok. Update the fixture or
expectation in the test command so it matches correct -z semantics, preserving
the existing checks for -d, -f, and -e.
In `@tests/t100_run_validation.sh`:
- Around line 37-48: Update the non-clang compiler enforcement test around the
gcc availability check to run unconditionally using a guaranteed non-clang
executable such as /bin/sh. Remove the environment-dependent skip, while
preserving the non-zero status assertion and “not a clang-family compiler”
message validation; if no suitable executable can be selected, fail the test
explicitly.
In `@tests/t101_run_basic.sh`:
- Around line 66-71: Strengthen the second invocation in the cache reuse test
around the mcsh command so it runs with verbose output and asserts that no
compile or link actions are emitted, while preserving the existing status and
“hello” output checks. Verify cached artifacts are reused rather than merely
confirming repeated execution correctness.
- Around line 80-85: Extend the run argument-passthrough test around the
existing status/output assertion to also invoke run with the `--` separator
before `one two`. Assert that this dedicated parsing path returns success and
produces exactly `one two`, using the same failure reporting and exit behavior
as the existing direct-forwarding check.
---
Outside diff comments:
In `@sh.init.c`:
- Around line 43-153: Restore the platform-guarded builtin registrations removed
from the builtin table, including att, bye, getspath, getxvers, migrate, and
universe, using their original conditional compilation guards and handlers. Keep
the new run entry and existing alphabetical ordering intact, ensuring each
platform retains its supported commands.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: ASSERTIVE
Plan: Pro Plus
Run ID: 64a83388-2554-4b51-91d0-ef216852d958
📒 Files selected for processing (47)
ISSUES.mdMakefile.inPLAN.mdREADME.mdSYNTAX_AND_SCRIPTING.mddocs/c-commands-spec.mddocs/c-workflow-analysis.mddocs/compiler/ARCHITECTURE.mddocs/compiler/CACHE-DESIGN.mddocs/compiler/CLI-SPEC.mddocs/compiler/DATA-MODEL.mddocs/compiler/DECISIONS.mddocs/compiler/DEPENDENCY-DISCOVERY.mddocs/compiler/INDEX.mddocs/compiler/MASTER-PLAN.mddocs/compiler/MILESTONES.mddocs/compiler/PIPELINES.mddocs/compiler/RISKS.mddocs/compiler/TEST-PLAN.mddocs/compiler/TODO.mdfix_truncation.patchsh.csh.cworkflow.csh.decls.hsh.dir.csh.func.csh.init.csh.sem.cstrncpy_analysis.mdtc.prompt.ctc.prompt.c.origtcsh.man.intest2test3tests/t005_cd_stack.shtests/t015_dotmcshrc_ls_colors.shtests/t017_cd_minus_n.shtests/t018_syntax_highlight.shtests/t019_git_prompt.shtests/t020_predict.shtests/t021_onintr.shtests/t022_function_args.shtests/t023_filetest_ops.shtests/t024_variable_modifiers.shtests/t025_switch.shtests/t100_run_validation.shtests/t101_run_basic.sh
💤 Files with no reviewable changes (3)
- strncpy_analysis.md
- tc.prompt.c.orig
- fix_truncation.patch
🛑 Comments failed to post (10)
docs/c-commands-spec.md (1)
1-3: 📐 Maintainability & Code Quality | 🟠 Major | ⚡ Quick win
Update the documentation from “design-only” to the delivered P1 scope.
The PR registers and implements
run; these documents currently tell users and contributors that no runtime exists, while CLI documentation also implies that unimplementedcompileandbuildare available.
docs/c-commands-spec.md#L1-L3: describerunas implemented and retaincompile/buildas planned.docs/c-workflow-analysis.md#L1-L3: distinguish completed P1 behavior from future phases.docs/compiler/ARCHITECTURE.md#L1-L4: update the status and remaining “future/not implemented” integration language.docs/compiler/CACHE-DESIGN.md#L1-L4: distinguish the implemented cache layout from future index/metadata features.docs/compiler/CLI-SPEC.md#L1-L17: markcompile/buildsyntax as planned; document only supportedrunoptions.docs/compiler/DATA-MODEL.md#L1-L4: state that this is the target model beyond the initial implementation.docs/compiler/DEPENDENCY-DISCOVERY.md#L1-L4: identify dependency scanning as future work rather than claiming no workflow exists.docs/compiler/INDEX.md#L1-L5: remove the claim that no runtime is implemented.📍 Affects 8 files
docs/c-commands-spec.md#L1-L3(this comment)docs/c-workflow-analysis.md#L1-L3docs/compiler/ARCHITECTURE.md#L1-L4docs/compiler/CACHE-DESIGN.md#L1-L4docs/compiler/CLI-SPEC.md#L1-L17docs/compiler/DATA-MODEL.md#L1-L4docs/compiler/DEPENDENCY-DISCOVERY.md#L1-L4docs/compiler/INDEX.md#L1-L5🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@docs/c-commands-spec.md` around lines 1 - 3, Update the documentation to reflect delivered P1 support: in docs/c-commands-spec.md (lines 1-3) describe run as implemented while keeping compile/build planned; in docs/c-workflow-analysis.md (lines 1-3) distinguish completed P1 behavior from future phases; in docs/compiler/ARCHITECTURE.md (lines 1-4) update status and remaining integration work; in docs/compiler/CACHE-DESIGN.md (lines 1-4) distinguish the implemented cache layout from future index/metadata features; in docs/compiler/CLI-SPEC.md (lines 1-17) mark compile/build syntax as planned and document only supported run options; in docs/compiler/DATA-MODEL.md (lines 1-4) identify the model as a future target beyond the initial implementation; in docs/compiler/DEPENDENCY-DISCOVERY.md (lines 1-4) describe dependency scanning as future work; and in docs/compiler/INDEX.md (lines 1-5) remove the claim that no runtime is implemented.docs/compiler/CACHE-DESIGN.md (1)
55-77: 🗄️ Data Integrity & Integration | 🟠 Major | ⚡ Quick win
Preserve ordered inputs in cache keys.
Linker flag order and
-Isearch order affect behavior; XOR also loses multiplicity. Hash a length-delimited, ordered sequence of object keys and effective linker/compiler argv instead of XORing objects or sorting order-sensitive flags. Otherwise distinct builds can reuse the wrong artifact.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@docs/compiler/CACHE-DESIGN.md` around lines 55 - 77, Update the cache-key design in the documented key construction and normalization section to preserve order and multiplicity: replace XOR of object keys with a length-delimited ordered sequence hash, and hash the effective linker/compiler argv in its original order. Remove sorting of order-sensitive linker flags and canonicalized -I paths, while retaining only normalization that does not change argument order or meaning.docs/compiler/DECISIONS.md (1)
223-227: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win
Align ADR-006 with the implemented execution model.
cw_run_execute_stage()returns a child status todorun(), which then updates shell status and cleans up. Document the fork-and-wait behavior; directexecvphere would not return to that pipeline.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@docs/compiler/DECISIONS.md` around lines 223 - 227, Update the ADR-006 execution-model documentation around the `run` behavior to match the implementation: `cw_run_execute_stage()` runs the stage in a child, returns its status to `dorun()`, and `dorun()` updates shell status and performs cleanup. Remove or correct the recommendation to use direct `execvp` so the documented model preserves fork-and-wait behavior and returns control to the pipeline.docs/compiler/DEPENDENCY-DISCOVERY.md (1)
145-158: 🗄️ Data Integrity & Integration | 🟠 Major | ⚡ Quick win
Do not treat mtime/size gating as cache-safe.
A header can change while retaining its size and timestamp, causing a stale cache hit. Always hash cache-key inputs for correctness, or explicitly define this as a best-effort optimization with a verification path.
🧰 Tools
🪛 markdownlint-cli2 (0.23.0)
[warning] 149-149: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@docs/compiler/DEPENDENCY-DISCOVERY.md` around lines 145 - 158, Revise the “mtime-gated hashing” section to remove the claim that matching mtime and size safely validates content_hash. Require hashing of cache-key inputs for correctness, or explicitly document the optimization as best-effort and add a verification path before accepting the cached hash.docs/compiler/MASTER-PLAN.md (1)
3-4: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win
Synchronize compiler planning documents with the implemented PR.
docs/compiler/MASTER-PLAN.md#L3-L4: update the “no runtime implemented” status and corresponding P1 checklist.docs/compiler/MILESTONES.md#L3-L4: update the planning-only status and mark the shipped M1 work accurately.📍 Affects 2 files
docs/compiler/MASTER-PLAN.md#L3-L4(this comment)docs/compiler/MILESTONES.md#L3-L4🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@docs/compiler/MASTER-PLAN.md` around lines 3 - 4, Synchronize the compiler planning documents with the implemented PR: in docs/compiler/MASTER-PLAN.md at lines 3-4, replace the no-runtime status and update the corresponding P1 checklist; in docs/compiler/MILESTONES.md at lines 3-4, replace the planning-only status and accurately mark the shipped M1 work.docs/compiler/PIPELINES.md (1)
291-303: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win
Documented "preferred" execute path doesn't match the shipped implementation.
This doc states the "Preferred path" is
execvp(binary_path, program_args)replacing the mcsh process directly. However,sh.cworkflow.c'scw_execute_binary/cw_run_execute_stagealwaysfork()s and thenwaitpid()s in the parent — matching RISKS.md R-T02's Contingency ("Always fork before exec indorun"), not this section's stated preference. Worth updating this doc to describe the actually-shipped fork+exec+waitpid design as the implemented approach, since a future reader could otherwise "fix" the code to match this doc and reintroduce the exact hazard R-T02 warns about (replacing the shell process itself in a non-forked builtin context).🧰 Tools
🪛 LanguageTool
[style] ~296-~296: Consider an alternative for the overused word “exactly”.
Context: ... the full environment. - Exit code is exactly the binary's exit code. - This is the...(EXACTLY_PRECISELY)
[grammar] ~300-~300: Ensure spelling is correct
Context: ...mple foregroundruninvocation, the builtin must arrange exec without an extra in...(QB_NEW_EN_ORTHOGRAPHY_ERROR_IDS_1)
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@docs/compiler/PIPELINES.md` around lines 291 - 303, Update the Stage 4 “Execute” section to document the shipped fork+exec+waitpid behavior in cw_execute_binary/cw_run_execute_stage rather than direct execvp replacement. Describe the forked child executing the binary, the parent waiting and propagating its status, and retain the exec-failure exit code 127; remove claims that run replaces the mcsh process directly or requires avoiding an intermediate shell process.ISSUES.md (1)
9-21: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win
Planning docs weren't updated to reflect the
runimplementation shipped in this same PR. Four documents still describe the P1run file.cmilestone as not-yet-implemented / open, butsh.cworkflow.c+sh.decls.hin this PR already ship a workingdorun()builtin (file and directory targets,--clean,-v,--passthrough), wired into the builtin table. Shared root cause: the compiler-planning doc set wasn't refreshed alongside the implementation commit.
ISSUES.md#L9-L21: move CW-001 from "Open" to a "Resolved" entry (it already exceeds the stated MVP scope by also supporting directory targets).PLAN.md#L48-L57: update "Immediate next milestone: M1" and the "No runtime implementation yet" status line to reflect thatrunis implemented (and note directory/project support was pulled forward from M3).docs/compiler/PIPELINES.md#L1-L6: update the "Status: planning / documentation only. No implementation exists yet." banner — at minimum for therunpipeline section, since it's now implemented.docs/compiler/TODO.md#L28-L83: check off the P1 boxes that are demonstrably done (sh.cworkflow.ccreated,dorun()implemented,sh.decls.h/sh.init.cwiring done,tests/t100_run_validation.sh/tests/t101_run_basic.shwritten).📍 Affects 4 files
ISSUES.md#L9-L21(this comment)PLAN.md#L48-L57docs/compiler/PIPELINES.md#L1-L6docs/compiler/TODO.md#L28-L83🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@ISSUES.md` around lines 9 - 21, Refresh the compiler planning documentation to match the shipped dorun() implementation: in ISSUES.md lines 9-21, move CW-001 from Open to Resolved and note its directory-target support; in PLAN.md lines 48-57, mark M1/run as implemented and note directory/project support moved forward from M3; in docs/compiler/PIPELINES.md lines 1-6, remove or revise the planning-only/no-implementation status for the run pipeline; and in docs/compiler/TODO.md lines 28-83, check off the completed P1 items for sh.cworkflow.c, dorun(), sh.decls.h/sh.init.c wiring, and the run tests.tcsh.man.in (2)
8483-8496: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win
Align
%Goperation-state documentation withgit_get_info.
REBASE_HEADis not used to produceREBASING;tc.prompt.cinstead detectsrebase-apply/rebasing,rebase-apply(AM),REVERT_HEAD, andBISECT_LOG. Document the actual paths and states so prompt users are not misled.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@tcsh.man.in` around lines 8483 - 8496, Update the %G operation-state documentation to match the detection logic in git_get_info: remove REBASE_HEAD and rebase-merge/head-name, and document rebase-apply/rebasing, rebase-apply (AM), REVERT_HEAD, and BISECT_LOG with their corresponding states.
8909-8913: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win
Correct the syntax-category count.
The text says 12 categories but lists only 10. Either enumerate the missing categories or remove the count.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@tcsh.man.in` around lines 8909 - 8913, Correct the category-count statement in the compile-time default-colors documentation near the listed token categories: either update the enumeration to include all 12 categories or remove the unsupported numeric count, while keeping the existing category descriptions accurate.tests/t023_filetest_ops.sh (1)
4-12: 🎯 Functional Correctness | 🟠 Major | ⚡ Quick win
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash set -euo pipefail echo "== candidate file ==" if [ -f tests/t023_filetest_ops.sh ]; then wc -l tests/t023_filetest_ops.sh cat -n tests/t023_filetest_ops.sh else echo "missing tests/t023_filetest_ops.sh" fd -a 't023_filetest_ops\.sh' . || true fi echo echo "== mktemp files in repo ==" rg -n '\$\(mktemp\)|! -z|z_ok|-z' tests || true echo echo "== shell availability and semantics probe ==" for shell in csh tcsh sh bash zsh; do if command -v "$shell" >/dev/null 2>&1; then echo "--- $shell ---" else echo "--- $shell: missing ---" continue fi shell -c 'tmpfile="$(mktemp)"; trap "rm -f \"$tmpfile\"" EXIT; printf "created %s\\n" "$tmpfile"; if (-z "$tmpfile") echo csh_zero_pred; if (! -z "$tmpfile") echo csh_neg_not_zero_pred' || (false; sed -n '1,80p' /tmp/test_log 2>/dev/null || true) done echo echo "== csh zero predicate direct probe ==" tmpfile="$(mktemp)" || exit 77 trap 'rm -f "$tmpfile"' EXIT csh -f -c 'if (-z "$tmpfile") echo zero; if (! -z "$tmpfile") echo not_zero' "$tmpfile" || true printf 'x' > "$tmpfile" || exit 77 csh -f -c 'if (-z "$tmpfile") echo zero; if (! -z "$tmpfile") echo not_zero' "$tmpfile" || trueRepository: orpheus497/mcsh
Length of output: 1787
🌐 Web query:
tcsh -z file inquiry operator zero size documentation💡 Result:
In tcsh, the file inquiry operator for checking if a file has zero size is -z [1][2]. When used in an expression, such as an if statement, -z returns true if the specified file exists and has a size of zero bytes [1][3][2]. Conversely, the -s operator is used to test if a file has a non-zero size (i.e., it exists and is larger than zero bytes) [1][2]. Example usage: if ( -z filename) then echo "The file is empty." endif Additionally, if you need to retrieve the actual size of a file in bytes, you can use the -Z operator [2][4]. Unlike the boolean -z operator, -Z returns the size of the file as an integer [2][4]. These file inquiry operators can also be evaluated using the built-in filetest command [2][5][6].
Citations:
- 1: http://www.physics.ntua.gr/konstant/ph150/Lectures/04/tcsh.pdf
- 2: https://manpages.ubuntu.com/manpages/bionic/man1/tcsh.1.html
- 3: https://www.oocities.org/gurucoder/Tools/sys/tcsh_syntax.txt
- 4: https://learnxinyminutes.com/tcsh/
- 5: https://web.pa.msu.edu/reference/tcsh-web/Builtin_commands.html
- 6: https://linux.die.net/man/1/tcsh
Fix the
-ztest fixture or expectation.
tmpfileis empty, so C-shell-zis true;! -z "$tmpfile"will not printz_ok, causing this test to fail against correct-zsemantics.Suggested fixture fix
tmpfile=$(mktemp) || exit 77 trap 'rm -rf "$tmpdir" "$tmpfile"' EXIT +printf 'x' > "$tmpfile" || exit 77📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.tmpdir=$(mktemp -d) || exit 77 tmpfile=$(mktemp) || exit 77 trap 'rm -rf "$tmpdir" "$tmpfile"' EXIT printf 'x' > "$tmpfile" || exit 77 out=$("$MCSH" -f -c " if (-d \"$tmpdir\") echo d_ok if (-f \"$tmpfile\") echo f_ok if (-e \"$tmpfile\") echo e_ok if (! -z \"$tmpfile\") echo z_ok🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@tests/t023_filetest_ops.sh` around lines 4 - 12, The test fixture around the C-shell -z check currently uses an empty tmpfile while expecting ! -z to print z_ok. Update the fixture or expectation in the test command so it matches correct -z semantics, preserving the existing checks for -d, -f, and -e.
|
@copilot Verify each finding against current code. Fix only still-valid issues, skip the Inline comments:
In
In
In
In
In
In
In
In
In
In
In
Outside diff comments:
Verify each finding against current code. Fix only still-valid issues, skip the Inline comments:
In
In
In
In
In
In
In
In
In
In
In
Outside diff comments:
|
... All still-valid findings addressed in commit Code change ( Documentation updates
|
There was a problem hiding this comment.
Actionable comments posted: 8
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (2)
docs/compiler/PIPELINES.md (1)
267-272: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick winAlign the run validation docs with the actual guards.
cw_run_validate_requestonly rejects.mcshfiles and then rejects non-.cfiles; it does not read the first line or check a shebang. Limit this sentence to the.mcshguard, or add the file-open/shebang check in the same function.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@docs/compiler/PIPELINES.md` around lines 267 - 272, Update the run validation documentation to state that cw_run_validate_request rejects arguments ending in .mcsh, then rejects non-.c files; remove the unsupported shell-script/shebang detection claim unless the corresponding check is added to cw_run_validate_request.docs/compiler/MASTER-PLAN.md (1)
91-91: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick winRecord the complete toolchain selection contract.
The completed task mentions only clang lookup through
PATH. The shippedruncontract also validates clang-family compilers and supports$mcsh_cc. Include both paths in this milestone entry.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@docs/compiler/MASTER-PLAN.md` at line 91, Update the completed P1 toolchain entry in MASTER-PLAN.md to document the full run contract: clang-family compiler validation and compiler selection through both PATH lookup and the $mcsh_cc environment variable. Preserve the existing BSD-licensing and completion status.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@docs/compiler/ARCHITECTURE.md`:
- Around line 3-4: Update the future integration example in Section 3.1 of
ARCHITECTURE.md to remove the stale dorun entry, leaving only the compile and
build placeholders while preserving the shipped run status.
In `@docs/compiler/CACHE-DESIGN.md`:
- Around line 3-4: Update the P1 cache design documentation to describe the
identity generated by cw_run_prepare_state(): sorted .c/.h paths and contents
form project_hash, which combines with cc_path for object and binary cache
paths. Define changes to compiler flags, include search paths, link flags,
source order, or object hash order as invalidating directory-scoped cached
artifacts, and revise the object/binary key contracts to remove unsupported
dep_hash, profile_hash, and artifact-metadata-order components.
In `@docs/compiler/CLI-SPEC.md`:
- Line 11: Add a language tag such as text to the changed Markdown fence at
docs/compiler/CLI-SPEC.md lines 11-11 and docs/compiler/DEPENDENCY-DISCOVERY.md
lines 151-151, preserving the existing fenced content.
- Line 14: Update the run command syntax overview to represent both supported
argument forms: direct optional program arguments and arguments following an
optional `--` separator. Keep the existing run command structure and target
notation unchanged.
In `@docs/compiler/INDEX.md`:
- Around line 3-4: Update the quick-reference TODO entry in the compiler
documentation so it no longer presents the P1 queue as the immediate next task
now that run is shipped; either point it to post-P1 work or clearly label the P1
checklist as completed, while preserving the existing P2 and P3 planning
references.
In `@docs/compiler/MILESTONES.md`:
- Around line 34-54: Update the M1 title and objective to explicitly include
directory targets as shipped P1 behavior, and revise the dorun() validation task
so it permits supported directory targets while still rejecting invalid inputs
such as .mcsh files. Keep the existing execution task and directory-target
support wording consistent with this contract.
In `@PLAN.md`:
- Around line 50-55: Complete the P1 documentation status migration: in PLAN.md
lines 50-55, update the Phase 2 status to show run as implemented and only
compile and build as planned; in docs/c-commands-spec.md line 3, mark run syntax
and behavior as delivered, including directory targets and supported options; in
docs/c-workflow-analysis.md lines 3-4, replace remaining P1 implementation
instructions with delivered behavior or historical design context.
In `@sh.cworkflow.c`:
- Around line 1001-1006: Update the exec error-pipe handling around the child
write and parent read to retry operations interrupted by EINTR and ensure the
full exec_errno object is transferred, including handling partial writes and
reads. When either pipe operation fails or cannot complete, propagate a distinct
internal-error result rather than allowing the failure to be interpreted as
normal exit status 127; preserve the existing execv failure diagnostic path for
successfully transferred errno data.
---
Outside diff comments:
In `@docs/compiler/MASTER-PLAN.md`:
- Line 91: Update the completed P1 toolchain entry in MASTER-PLAN.md to document
the full run contract: clang-family compiler validation and compiler selection
through both PATH lookup and the $mcsh_cc environment variable. Preserve the
existing BSD-licensing and completion status.
In `@docs/compiler/PIPELINES.md`:
- Around line 267-272: Update the run validation documentation to state that
cw_run_validate_request rejects arguments ending in .mcsh, then rejects non-.c
files; remove the unsupported shell-script/shebang detection claim unless the
corresponding check is added to cw_run_validate_request.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: ASSERTIVE
Plan: Pro Plus
Run ID: d6309327-02af-4c02-84d3-06e1bb984997
📒 Files selected for processing (16)
ISSUES.mdPLAN.mddocs/c-commands-spec.mddocs/c-workflow-analysis.mddocs/compiler/ARCHITECTURE.mddocs/compiler/CACHE-DESIGN.mddocs/compiler/CLI-SPEC.mddocs/compiler/DATA-MODEL.mddocs/compiler/DECISIONS.mddocs/compiler/DEPENDENCY-DISCOVERY.mddocs/compiler/INDEX.mddocs/compiler/MASTER-PLAN.mddocs/compiler/MILESTONES.mddocs/compiler/PIPELINES.mddocs/compiler/TODO.mdsh.cworkflow.c
|
@copilot Verify each finding against current code. Fix only still-valid issues, skip the Inline comments:
In
In
In
In
In
In
Outside diff comments:
In
|
sh.cworkflow.c: - cw_execute_binary: child write now retries on EINTR and accumulates partial writes to transfer the full exec_errno value - cw_execute_binary: parent read now retries on EINTR, accumulates partial reads, and returns -3 on read error or partial transfer (distinct from -2 exec failure and -1 fork/pipe-setup failure) - cw_run_execute_stage: handle -3 with "run: internal error: exec pipe failed" diagnostic Documentation: - ARCHITECTURE.md §3.1: remove stale dorun from planned-only example - CACHE-DESIGN.md §2: replace planned P2 key design with actual P1 project_hash/object/binary key contracts; move P2/P3 planned keys to new §2.4/2.5 - CLI-SPEC.md: add `text` language tag to syntax fence; show both run arg forms (with and without -- separator) - DEPENDENCY-DISCOVERY.md §4.3: add `text` language tag to fence - INDEX.md: update "Immediate next TODO" to M2 (P1 is done) - MILESTONES.md: M1 title/objective include directory targets; fix validation task to permit directories while rejecting .mcsh - PLAN.md: update Phase 2 "Current status" to reflect M1 shipped - MASTER-PLAN.md: expand P0 toolchain entry to include $mcsh_cc and clang-family validation; check off all P1 items - PIPELINES.md Stage 1: remove unsupported shebang-detection claim - docs/c-commands-spec.md: add directory targets to status line - docs/c-workflow-analysis.md: Phase 1 section describes delivered behavior; stale "next milestone" replaced with historical note; integration-points section reflects shipped state
... All still-valid findings addressed in commit
Documentation (all findings valid, fixed)
|
There was a problem hiding this comment.
Actionable comments posted: 7
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (5)
docs/compiler/DEPENDENCY-DISCOVERY.md (1)
3-5: 🗄️ Data Integrity & Integration | 🟠 Major | ⚡ Quick winSeparate P1 invalidation from planned dependency tracking.
The status says P1 uses source-only content-hash keys and full header tracking is future P2 work. The invalidation rules still state that
rundetects header changes, deleted headers, and new dependency edges. Mark those rules as planned P2/P3 behavior, or add separate P1 and future-phase sections.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@docs/compiler/DEPENDENCY-DISCOVERY.md` around lines 3 - 5, Update the invalidation rules in the dependency-discovery documentation to distinguish current P1 `run` behavior from planned dependency tracking: remove or relabel claims that `run` detects header changes, deleted headers, or new dependency edges as future P2/P3 behavior, while preserving the source-content-hash-only P1 cache-key description.docs/compiler/CACHE-DESIGN.md (1)
3-4: 🗄️ Data Integrity & Integration | 🟡 Minor | ⚡ Quick winScope the artifact metadata sections to planned phases.
The status identifies
index.dband sidecar metadata as planned work. Later artifact and lookup sections describe.metafiles and metadata validation as current cache behavior. Mark those sections as planned P2/P3 design, or document the actual P1 path-only lookup.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@docs/compiler/CACHE-DESIGN.md` around lines 3 - 4, The documentation currently presents artifact `.meta` files and metadata validation as implemented despite the status limiting P1 to object and binary cache paths. Update the later artifact and lookup sections to label `index.db`, sidecar metadata, and related validation as planned P2/P3 work, or revise them to accurately describe the implemented P1 path-only lookup.docs/compiler/CLI-SPEC.md (1)
150-162: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick winSynchronize the option matrix with the implemented
runsyntax.This section removes
-p,-I,-D, and-gfromrun, but the option matrix still marks all four options as supported. Change those matrix entries to unsupported, or restore matching implementation and tests.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@docs/compiler/CLI-SPEC.md` around lines 150 - 162, Update the option matrix in the run command documentation to mark -p, -I, -D, and -g as unsupported, matching the implemented run syntax; do not leave these options listed as supported unless their implementation and tests are restored.docs/compiler/PIPELINES.md (1)
301-307: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick winDocument every exec-pipe result and include
runstatus255.The documented
runoutcome only lists127for allexecvfailures and shows program exit statuses only through0..254. Add the internal-error path for partial/failing pipe transfer, including its diagnosticrun: internal error: exec pipe failedandexit code 127, and extend the executed-binary status range to include255.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@docs/compiler/PIPELINES.md` around lines 301 - 307, The PIPELINES.md “run” outcome documentation should cover every exec-pipe result: document the partial or failed pipe transfer as `run: internal error: exec pipe failed` with exit code 127, and extend the executed binary’s propagated exit-status range from 0..254 to include 255.docs/compiler/MILESTONES.md (1)
50-53: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick winDocument the same argument-forwarding contract in both files.
The shipped interface accepts direct arguments and an optional
--separator. It preserves the tokens after the target as program arguments.
docs/compiler/MILESTONES.md#L50-L53: replace “with--args” with direct-or-separator argument forwarding.docs/compiler/PIPELINES.md#L294-L300: state that the child receives the original argument order and multiplicity, whether or not--is present.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@docs/compiler/MILESTONES.md` around lines 50 - 53, Update the dorun argument-forwarding documentation in docs/compiler/MILESTONES.md lines 50-53 to describe direct arguments or an optional -- separator instead of only “with -- args”; update docs/compiler/PIPELINES.md lines 294-300 to state that the child receives tokens after the target in their original order and multiplicity, regardless of whether -- is present.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@docs/compiler/ARCHITECTURE.md`:
- Around line 126-131: Update docs/compiler/ARCHITECTURE.md lines 126-131 to
remove dorun() from future declaration/implementation examples or clearly mark
those sections as current state; update docs/compiler/MILESTONES.md line 41 to
replace “dorun() stub” with wording that reflects the completed implementation.
In `@docs/compiler/CACHE-DESIGN.md`:
- Line 41: Add a suitable language tag, such as text, to each new pseudocode
fenced block in CACHE-DESIGN.md, including the blocks at the referenced
locations, while leaving their pseudocode content unchanged.
- Around line 73-89: Update the P1 binary cache documentation around the “run —
binary cache key” section so invalidation claims match the displayed key: only
source/header changes represented by project_hash and compiler identity
represented by cc_hash should be described as P1 invalidators. Describe compiler
flags, include paths, linker flags, source order, and object order as outside
the P1 cache identity and directory-cache limitations, unless the key,
implementation, and tests are expanded accordingly.
- Around line 127-134: Update the “Normalization before hashing” section to
preserve repeated compiler flags and remove the instruction to eliminate
duplicates such as repeated -O options. Document that normalization must retain
argument order and multiplicity, while continuing to normalize path separators
and leave include paths and linker flags unsorted.
- Around line 35-52: Update the project_hash formula in the “2.1 P1 run —
project hash” section to explicitly include each fingerprint’s path, content,
and delimiter, and list source position if it is an actual input; otherwise
remove unsupported position claims. Ensure the surrounding prose and formula
consistently describe the inputs needed to distinguish different project
layouts.
In `@docs/compiler/MASTER-PLAN.md`:
- Around line 98-101: Update the command contract in the P1 section of
MASTER-PLAN.md to accept both individual C files and directory/project targets,
changing the documented run syntax from file-only to a form such as run
<file.c|target> [...]. Keep the existing compilation and execution behavior
unchanged.
- Around line 102-111: Unify compiler validation so every selected compiler must
pass clang-family validation. In docs/compiler/MASTER-PLAN.md lines 102-111,
clarify that $mcsh_cc or the compiler selected from PATH is mandatory
clang-family validated; make the corresponding discovery step in
docs/compiler/MILESTONES.md lines 45-46 explicitly require validation rather
than merely preferring clang or falling back to cc.
---
Outside diff comments:
In `@docs/compiler/CACHE-DESIGN.md`:
- Around line 3-4: The documentation currently presents artifact `.meta` files
and metadata validation as implemented despite the status limiting P1 to object
and binary cache paths. Update the later artifact and lookup sections to label
`index.db`, sidecar metadata, and related validation as planned P2/P3 work, or
revise them to accurately describe the implemented P1 path-only lookup.
In `@docs/compiler/CLI-SPEC.md`:
- Around line 150-162: Update the option matrix in the run command documentation
to mark -p, -I, -D, and -g as unsupported, matching the implemented run syntax;
do not leave these options listed as supported unless their implementation and
tests are restored.
In `@docs/compiler/DEPENDENCY-DISCOVERY.md`:
- Around line 3-5: Update the invalidation rules in the dependency-discovery
documentation to distinguish current P1 `run` behavior from planned dependency
tracking: remove or relabel claims that `run` detects header changes, deleted
headers, or new dependency edges as future P2/P3 behavior, while preserving the
source-content-hash-only P1 cache-key description.
In `@docs/compiler/MILESTONES.md`:
- Around line 50-53: Update the dorun argument-forwarding documentation in
docs/compiler/MILESTONES.md lines 50-53 to describe direct arguments or an
optional -- separator instead of only “with -- args”; update
docs/compiler/PIPELINES.md lines 294-300 to state that the child receives tokens
after the target in their original order and multiplicity, regardless of whether
-- is present.
In `@docs/compiler/PIPELINES.md`:
- Around line 301-307: The PIPELINES.md “run” outcome documentation should cover
every exec-pipe result: document the partial or failed pipe transfer as `run:
internal error: exec pipe failed` with exit code 127, and extend the executed
binary’s propagated exit-status range from 0..254 to include 255.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: ASSERTIVE
Plan: Pro Plus
Run ID: b93b01fe-95ce-4540-87da-be2bba37e5f2
📒 Files selected for processing (12)
PLAN.mddocs/c-commands-spec.mddocs/c-workflow-analysis.mddocs/compiler/ARCHITECTURE.mddocs/compiler/CACHE-DESIGN.mddocs/compiler/CLI-SPEC.mddocs/compiler/DEPENDENCY-DISCOVERY.mddocs/compiler/INDEX.mddocs/compiler/MASTER-PLAN.mddocs/compiler/MILESTONES.mddocs/compiler/PIPELINES.mdsh.cworkflow.c
| ### P1 — `run file.c` MVP ✓ shipped | ||
|
|
||
| **Goal:** `run hello.c` compiles hello.c (if needed), executes it, returns exit status. | ||
|
|
There was a problem hiding this comment.
🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win
Update the immutable contract for directory targets.
The completed P1 section documents directory/project targets, but the command contract still lists run file.c [...] only. Change that contract to include a directory target, such as run <file.c|target> [...].
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@docs/compiler/MASTER-PLAN.md` around lines 98 - 101, Update the command
contract in the P1 section of MASTER-PLAN.md to accept both individual C files
and directory/project targets, changing the documented run syntax from file-only
to a form such as run <file.c|target> [...]. Keep the existing compilation and
execution behavior unchanged.
| - [x] Implement `dorun()` in new `sh.cworkflow.c` | ||
| - [x] Register `"run"` in `bfunc[]` (`sh.init.c`) in sorted position | ||
| - [x] Add `extern void dorun(Char **, struct command *)` in `sh.decls.h` | ||
| - [x] Input validation: accept `.c` files and directories; reject `.mcsh` with guidance error | ||
| - [x] Cache key: SHA-256 of source content + compiler identity (project_hash + cc_hash) | ||
| - [x] Cache directory: `~/.mcsh_cache/cworkflow/` | ||
| - [x] Compile with discovered system cc (clang-family preferred; `$mcsh_cc` or PATH) | ||
| - [x] Execute compiled binary; propagate child exit status | ||
| - [x] Basic diagnostics: compile errors printed to stderr with file:line format | ||
| - [x] Directory/project target support |
There was a problem hiding this comment.
🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win
Use one compiler-validation contract.
The implementation requires a clang-family compiler, but the documents use “preferred” and “clang → cc” wording that can imply non-clang compilers are accepted.
docs/compiler/MASTER-PLAN.md#L102-L111: state that$mcsh_ccor the PATH-selected compiler must pass clang-family validation.docs/compiler/MILESTONES.md#L45-L46: document discovery followed by mandatory clang-family validation.
📍 Affects 2 files
docs/compiler/MASTER-PLAN.md#L102-L111(this comment)docs/compiler/MILESTONES.md#L45-L46
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@docs/compiler/MASTER-PLAN.md` around lines 102 - 111, Unify compiler
validation so every selected compiler must pass clang-family validation. In
docs/compiler/MASTER-PLAN.md lines 102-111, clarify that $mcsh_cc or the
compiler selected from PATH is mandatory clang-family validated; make the
corresponding discovery step in docs/compiler/MILESTONES.md lines 45-46
explicitly require validation rather than merely preferring clang or falling
back to cc.
|
@copilot Verify each finding against current code. Fix only still-valid issues, skip the Inline comments:
In
In
Outside diff comments:
In
In
In
In
|
... All still-valid findings fixed in commit ARCHITECTURE.md §3.2/§3.3 — MILESTONES.md line 41 — "dorun() stub" → "dorun() implementation". CACHE-DESIGN.md §2.1 (project_hash formula) — formula updated to explicitly show CACHE-DESIGN.md §2.2 (object cache key prose) — removed "its position within the project" claim; prose now says "source path and content, the project-wide input set, and the compiler identity". CACHE-DESIGN.md §2.3 (binary cache key invalidation) — removed the inaccurate claim that compiler flags/include paths/linker flags invalidate the cache "because they alter project_hash or cc_hash" (they don't in P1). Replaced with an explicit note that those inputs are outside the P1 cache identity. CACHE-DESIGN.md §2.1/§2.2/§2.3 fenced blocks — all pseudocode fences now carry the CACHE-DESIGN.md §2.6 (normalization) — removed "Remove redundant flags (duplicate |
There was a problem hiding this comment.
Actionable comments posted: 1
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
docs/compiler/CACHE-DESIGN.md (1)
60-75: 🗄️ Data Integrity & Integration | 🟡 Minor | ⚡ Quick winDefine
cc_hashas a single fallback contract.The implementation hashes the compiler file first, then hashes the compiler path when file hashing fails. Document this as the exact
cc_hashinput and invalidation rule in the object key formula, binary key formula, and cache invalidators. A separate cached compiler binary means toolchain changes invalidate objects and binaries; a plain path hash would not.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@docs/compiler/CACHE-DESIGN.md` around lines 60 - 75, Update CACHE-DESIGN.md to define cc_hash consistently as the compiler binary-content hash when file hashing succeeds, with the compiler path hash as the sole fallback when it fails. Apply this exact contract to both object and binary key formulas and explicitly state that compiler changes invalidate cached objects and binaries, while path hashing is only the fallback.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@docs/compiler/CACHE-DESIGN.md`:
- Around line 37-50: Update cw_run_prepare_state() to canonicalize each source
and header path before adding it to the fingerprint list used by
cw_hash_file_set(). Ensure equivalent inputs such as file.c, ./file.c, and
/absolute/file.c use the same normalized path while preserving the existing
sorted fingerprint and content-hash behavior.
---
Outside diff comments:
In `@docs/compiler/CACHE-DESIGN.md`:
- Around line 60-75: Update CACHE-DESIGN.md to define cc_hash consistently as
the compiler binary-content hash when file hashing succeeds, with the compiler
path hash as the sole fallback when it fails. Apply this exact contract to both
object and binary key formulas and explicitly state that compiler changes
invalidate cached objects and binaries, while path hashing is only the fallback.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: ASSERTIVE
Plan: Pro Plus
Run ID: 13e9d258-7c8e-42fe-af05-895ede89ae55
📒 Files selected for processing (3)
docs/compiler/ARCHITECTURE.mddocs/compiler/CACHE-DESIGN.mddocs/compiler/MILESTONES.md
| `cw_run_prepare_state()` computes a **project_hash** that covers all source and | ||
| header inputs. Sorted `.c` and `.h` file paths are collected into the | ||
| fingerprint list; each entry contributes its path, a NUL delimiter, and its | ||
| content hash to the rolling digest: | ||
|
|
||
| ```text | ||
| project_hash = SHA-256( | ||
| path(fingerprints[0]) ← absolute path (NUL-terminated) | ||
| SHA-256(content(fingerprints[0])) ← content hash of that file | ||
| path(fingerprints[1]) | ||
| SHA-256(content(fingerprints[1])) | ||
| ... | ||
| path(fingerprints[N-1]) | ||
| SHA-256(content(fingerprints[N-1])) |
There was a problem hiding this comment.
🗄️ Data Integrity & Integration | 🟡 Minor | ⚡ Quick win
🧩 Analysis chain
🏁 Script executed:
#!/bin/sh
set -eu
rg -n -C 12 \
'cw_parse|target_path|realpath|cw_collect_project_files|cw_hash_file_set' \
sh.cworkflow.cRepository: orpheus497/mcsh
Length of output: 6613
🏁 Script executed:
#!/bin/sh
set -eu
sed -n '1139,1275p' sh.cworkflow.c
printf '\n--- find references to realpath/canonicalize/norm paths ---\n'
rg -n 'realpath|canon|normalize|target_path|fingerprints|cw_run_prepare_state|cw_hash_toolchain' sh.cworkflow.c docs/compiler/CACHE-DESIGN.mdRepository: orpheus497/mcsh
Length of output: 7094
Canonicalize paths before hashing.
run file.c saves req->target_path as file.c; stat accepts it, but cw_hash_file_set() hashes that literal fingerprint path. This can give different project_hash values for file.c, ./file.c, and /absolute/file.c. Normalize these paths before pushing to the fingerprint list.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@docs/compiler/CACHE-DESIGN.md` around lines 37 - 50, Update
cw_run_prepare_state() to canonicalize each source and header path before adding
it to the fingerprint list used by cw_hash_file_set(). Ensure equivalent inputs
such as file.c, ./file.c, and /absolute/file.c use the same normalized path
while preserving the existing sorted fingerprint and content-hash behavior.
|
@copilot Verify each finding against current code. Fix only still-valid issues, skip the Inline comments:
Outside diff comments:
|
…rify cc_hash in CACHE-DESIGN
... Both findings fixed in commit
|
Why
mcsh needs a first-class C workflow that is easier than juggling direct compiler and build-tool invocations. This branch documents the compiler/workflow design and lands the first executable slice: a
runcommand that drives compile, build, and execute in one path.What changed
docs/compiler/, plus related updates to top-level docs (README.md,PLAN.md,ISSUES.md,SYNTAX_AND_SCRIPTING.md) and C workflow analysis/spec docs.runin shell dispatch/build integration (sh.init.c,sh.decls.h,Makefile.in).sh.cworkflow.cwith a staged, data-oriented pipeline (parse -> validate -> prepare -> compile -> build -> execute).runnow supports both single C file targets and directory/project targets, with object and binary cache paths derived from source/toolchain hashes.$mcsh_ccvalidation) to keep runtime/tooling alignment intact.runvalidation and behavior tests (tests/t100_run_validation.sh,tests/t101_run_basic.sh) covering misuse handling, arg passthrough, exit code propagation, cache reuse, and directory builds.Notes for review
runpath intentionally focuses on streamlined compile/build/run behavior (--clean,-v, optional--separator) rather than broader profile/flag matrices; those are tracked in the planning docs.phase1; reviewing by commit will make the progression clearer.Summary by CodeRabbit
New Features
runcommand for compiling and executing C files or directory projects..mcshfiles, and supported compiler toolchains.Bug Fixes
cdtargets.Documentation
Tests