Hi Bruno, I think that the test-coreutils-1.sh test is missing --local-dir=gl. Was this intentional? From test-coreutils-1.err:
$ cat test-coreutils-1.err gnulib-tool: warning: module buffer-lcm doesn't exist gnulib-tool: warning: module cl-strtod doesn't exist gnulib-tool: warning: module cl-strtold doesn't exist ... It seems that gnulib-tool.sh will continue importing things after these error messages, which is why I think you may have missed it. gnulib-tool.py fails in an ugly manner: ... gnulib-tool: warning: file targetdir does not exist gnulib-tool: warning: file xdectoint does not exist gnulib-tool: warning: file xfts does not exist Traceback (most recent call last): File "/home/collin/.local/src/gnulib/pygnulib/main.py", line 1341, in <module> main() File "/home/collin/.local/src/gnulib/pygnulib/main.py", line 914, in main filetable, transformers = importer.prepare() ^^^^^^^^^^^^^^^^^^ File "/home/collin/.local/src/gnulib/pygnulib/GLImport.py", line 864, in prepare final_modules = self.moduletable.transitive_closure(base_modules) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/collin/.local/src/gnulib/pygnulib/GLModuleSystem.py", line 921, in transitive_closure raise TypeError('each module must be a GLModule instance, not %s' TypeError: each module must be a GLModule instance, not NoneType When passing --local-dir=gl to gnulib-tool.sh and gnulib-tool.py they seem to fail the test case similarly to each other, which seems like a good sign :). This is done in bootstrap.conf for coreutils: $ ./bootstrap --skip-po ./bootstrap: Bootstrapping from checked-out coreutils sources... ./bootstrap: autopoint --force ./bootstrap: /home/collin/.local/src/gnulib/gnulib-tool --no-changelog --aux-dir=build-aux --doc-base=doc --lib=libcoreutils --m4-base=m4/ --source-base=lib/ --tests-base=tests --local-dir=gl --tests-base=gnulib-tests --with-tests --symlink --makefile-name=gnulib.mk --automake-subdir --import ... Collin