tb put this in a bulk and saw textproc/catfish fail during fake.
Regardless of the previous python.port.mk diff I sent, I think the
module is subtly wrong wrt.
- MODPY_BUILDDEP: add python as a build dependency
- NO_BUILD: whether there is a build target at all
catfish sets NO_BUILD=Yes because it just packages existing files,
but that makes python.port.mk default to MODPY_BUILDDEP=No, which seems
logical but is not accurate.
catfish for example still uses MODPY_ADJ_FILES which uses python in
pre-configure and it still relies on python to execute the fake target.
So should the module's logic be adjusted or should ports like catfish
be more explicit?
The latter is simpler and safer, so the diff below reflects reality;
it should also prevent python from being junked in bulk builds, which
would make it fail, but my bulk/dpb foo is not good enough.
Index: Makefile
===================================================================
RCS file: /cvs/ports/textproc/catfish/Makefile,v
retrieving revision 1.56
diff -u -p -r1.56 Makefile
--- Makefile 23 Dec 2022 08:18:14 -0000 1.56
+++ Makefile 8 Jan 2023 15:06:02 -0000
@@ -17,6 +17,8 @@ MODULES= lang/python \
x11/xfce4
MODPY_ADJ_FILES=bin/catfish
NO_BUILD= Yes
+# NO_BUILD=Yes defaults this to No, but configure and fake need it
+MODPY_BUILDDEP= Yes
RUN_DEPENDS = devel/py-gobject3${MODPY_FLAVOR} \
devel/desktop-file-utils \
