Stop makefile from calling linkml validate with nonexistent config files #91

Merged

adina merged 2 commits from make-invalid-wildcard into main

2025-08-15 10:24:52 +00:00

msz commented

2025-08-13 12:00:27 +00:00

Owner

The checkvalidation (checkvalid / checkinvalid) target in the makefile globs for *(in)valid.cfg.yaml files. While #90 excluded checkinvalid from checkvalidation, this stops the (checkvalid / checkinvalid) targets from calling linkml validate when no files match the pattern instead.

Previously, if there were no *.invalid.cfg.yaml files, we still made it into the @for ex in $</*invalid.cfg.yaml loop (same for *valid) -- but with the glob expression verbatim, producing errors like this:

"Invalid value for '--config': File 'src/flat-data/unreleased/validation/*.invalid.cfg.yaml' does not exist."

It seems that make's wildcard function avoids the pitfall, while correctly expanding the globs when files are present.

The checkvalidation (checkvalid / checkinvalid) target in the makefile globs for `*(in)valid.cfg.yaml` files. While #90 excluded checkinvalid from checkvalidation, this stops the (checkvalid / checkinvalid) targets from calling `linkml validate` when no files match the pattern instead. Previously, if there were no *.invalid.cfg.yaml files, we still made it into the `@for ex in $</*invalid.cfg.yaml` loop (same for `*valid`) -- but with the glob expression verbatim, producing errors like this: "Invalid value for '--config': File 'src/flat-data/unreleased/validation/*.invalid.cfg.yaml' does not exist." It seems that make's [wildcard](https://www.gnu.org/software/make/manual/html_node/Wildcard-Function.html) function avoids the pitfall, while correctly expanding the globs when files are present.

msz added 2 commits

2025-08-13 12:00:28 +00:00

Fix makefile formatting dbd92ebea0

Removes three trailing spaces, which made my editor complain on save.

Stop makefile from calling linkml validate with config absent

Codespell / Check for spelling errors (pull_request) Successful in 25s

Details

Model checks / lint (pull_request) Successful in 1m42s

Details

Validate examples and verify unmodified conversion / lint (pull_request) Successful in 2m48s

Details

4abd9b940b

This stops the checkvalidation (checkvalid / checkinvalid) target in
the makefile, which globs for `*(in)valid.cfg.yaml` files, from
calling `linkml validate` when no files match the pattern.

Previously, if there were no *.invalid.cfg.yaml files, we still made
it into the `@for ex in $</*invalid.cfg.yaml` loop (same for `*valid`)
-- but with the glob expression verbatim, producing errors like this:

"Invalid value for '--config': File
'src/flat-data/unreleased/validation/*.invalid.cfg.yaml' does not
exist."

It seems that make's wildcard function avoids the pitfall, while
correctly expanding the globs when files are present.

This is an alternative to a solution presented in #90, which excluded
checkinvalid from checkvalidation.

https://www.gnu.org/software/make/manual/html_node/Wildcard-Function.html

adina referenced this pull request

2025-08-13 12:16:42 +00:00

ci: don't check invalid examples #90

msz referenced this pull request

2025-08-13 13:04:54 +00:00

ci: don't check invalid examples #90

adina referenced this pull request

2025-08-13 13:27:05 +00:00

ci: don't check invalid examples #90

msz changed title from ~~make-invalid-wildcard~~ to Stop makefile from calling linkml validate with nonexistent config files

2025-08-13 13:48:38 +00:00

mih commented

2025-08-14 11:27:46 +00:00

Owner

It would make sense to direct this PR to the datalad-concepts repo. This is the model source for all downstream efforts, whereas this is just one of the downstream usages.

msz commented

2025-08-14 16:50:29 +00:00

Author

Owner

@mih wrote in #91 (comment):

It would make sense to direct this PR to the datalad-concepts repo. This is the model source for all downstream efforts, whereas this is just one of the downstream usages.

Thanks, opened https://github.com/psychoinformatics-de/datalad-concepts/pull/318 now.

@mih wrote in https://hub.psychoinformatics.de/inm7/inm7-concepts/pulls/91#issuecomment-4500: > It would make sense to direct this PR to the datalad-concepts repo. This is the model source for all downstream efforts, whereas this is just one of the downstream usages. Thanks, opened https://github.com/psychoinformatics-de/datalad-concepts/pull/318 now.