Reply to my own thread!
I solved my problem by using QMAKE_EXTRA_COMPILER instead and treating the lint program as a separate compiler
mylint.output = ${QMAKE_FILE_BASE}.lint
mylint.input = SOURCES
mylint.CONFIG = no_link
mylint.commands = llint -os ${QMAKE_FILE_OUT} $${mylint_include_path} ${QMAKE_FILE_NAME}
QMAKE_EXTRA_COMPILERS += mylint
linttarget.target = lint
linttarget.depends = $$lint_targets # computed above (see original post)
QMAKE_EXTRA_TARGETS += linttarget
mylint.output = ${QMAKE_FILE_BASE}.lint
mylint.input = SOURCES
mylint.CONFIG = no_link
mylint.commands = llint -os ${QMAKE_FILE_OUT} $${mylint_include_path} ${QMAKE_FILE_NAME}
QMAKE_EXTRA_COMPILERS += mylint
linttarget.target = lint
linttarget.depends = $$lint_targets # computed above (see original post)
QMAKE_EXTRA_TARGETS += linttarget
To copy to clipboard, switch view to plain text mode
It ain't pretty, but it works. I now have an extra make target called lint that depends on fileone.lint filetwo.lint filethree.lint
Those in turn are generated by the extra compiler that invokes lint on each source.
I still would like to have a finer control on things like the include paths but without 3 or 4 weeks to dive into the guts of qmake, this works for me. Hopefully it helps someone else. 
-Mic
Bookmarks