COMP: Build the optional code paths in the per-PR workflow

[gdevenyi]

cmake-multi-platform.yml is the workflow that runs on every pull request, and it passes no options at all. With the defaults that means cifti/ and fsliolib/ are never compiled, and neither are the blocks behind FSLSTYLE, PIGZ and REJECT_COMPLEX:

Option	Default	Gates
USE_CIFTI_CODE	OFF	cifti/
USE_FSL_CODE	OFF	fsliolib/
FSLSTYLE_NAME_CONFLICTS	OFF	-DFSLSTYLE
FSLSTYLE_PIGZ_SUPPORT	OFF	-DPIGZ
FSLSTYLE_REJECT_COMPLEX	OFF	-DREJECT_COMPLEX

So a PR that changes any of those files collects four green checks that never built it. Several open PRs are in exactly that position — #54 and #74 change FSLSTYLE/REJECT_COMPLEX blocks that nothing in the repo compiles, and roughly eighteen touch cifti/ or fsliolib/.

#57 and #59 do set -DUSE_CIFTI_CODE=ON -DUSE_FSL_CODE=ON, but they add different workflows and leave this one alone, so the per-PR signal stays misleading until one of them merges — and neither covers the FSL parity defines.

The change. The matrix gains an options axis: default is what a consumer gets with no arguments, all turns the optional libraries and the FSL parity defines on. Both are kept, because the default build is what ships and a change can break it while the fuller one still compiles. Four jobs become eight, each about half a minute. expat is installed for the Linux all jobs.

Verified on master before writing it, with the test suite:

default                build OK, tests pass
+cifti +fsl            build OK, tests pass
+cifti +fsl +FSLSTYLE  build OK, tests pass,
                       doPigz/doPigz2 present in nifti2_io.c.o

COMPILE_NIFTIUNUSED_CODE is deliberately left out. It guards code the project itself labels unused, so compiling it in CI would commit to keeping it working; whether that code should exist at all is a separate question (it is where #41 lives).

Independent of #57 and #59, which touch other files. #28 edits the ctest line in this same file, one line away from nothing this PR touches.

🤖 Generated with Claude Code

[hjmjohnson]

Reviewed as part of a sweep over the open backlog. Merge as-is. Highest-value of the current CI group.

A nine-agent review of all 52 open PRs found this same hole from four independent directions: cifti/, fsliolib/, and the FSLSTYLE paths are compiled by no CI job on this repository. Concretely, PRs #54, #55, #56, #66, #73, and #74 all modify code that has never been built by CI here. This PR is what closes that.

Job count goes 4 -> 8 and all eight already pass, so enabling the optional paths surfaces no new breakage — the code compiles, it simply was not being checked.

What remains uncompiled after this

The all leg covers USE_CIFTI_CODE, USE_FSL_CODE, and FSLSTYLE. Still built by nothing:

• REJECT_COMPLEX — BUG: Fix the REJECT_COMPLEX path in nifti_image_read #74 modifies this path; it is plausibly never been compiled by anyone reviewing it
• FSLSTYLE_NAME_CONFLICTS — the configuration BUG: Fix the ambiguous-filename path in nifti_findhdrname #54's behavior change is confined to
• COMPILE_NIFTIUNUSED_CODE
• FSLSTYLE_PIGZ_SUPPORT — the popen half of the C11 question in COMP: Build as ISO C11 with no compiler extensions #38

Each is a one-line matrix addition if you want them, though I would not expand the matrix further in this PR.

Two notes, neither blocking

TEST_INSTALL is still OFF in this workflow. It is gated on BUILD_SHARED_LIBS (CMakeLists.txt:120), which defaults OFF (:51) and is not set here, so install_linking never registers in these jobs — see the discussion on #29. The install/export path (install_nifti_target, the generated NIFTIConfig.cmake, and the find_package(NIFTI ... CONFIG REQUIRED) in real_easy/stand_alone_app/) therefore remains untested on every pull request even after this lands. Separate change; flagging so the gap is not assumed closed.

Shares a file with #28 (.github/workflows/cmake-multi-platform.yml). Hunks do not overlap — #28 is the ctest line, this is the matrix and configure steps — so they merge cleanly, but whichever lands second is worth a base update first.

[hjmjohnson]

The commit messages in this range were rewritten to remove trailers that do not belong in permanent history: Co-Authored-By: naming an AI tool, and Claude-Session: URLs that resolve for nobody. Only messages changed — the tree at the tip of master is byte-identical, and author, committer, and dates are preserved.

This PR's commit on the rewritten master:

• b6531f011b COMP: Build the optional code paths in the per-PR workflow

The SHA recorded above by GitHub is from the pre-rewrite history and no longer resolves.