Paul Eggert wrote: > > #if USE_UNLOCKED_IO > > # include "unlocked-io.h" > > #endif > > > > to a simple > > > > #include "unlocked-io.h" > > > > then --avoid will do the trick as well. > > If you make that change to the source code, and then use --avoid, the file > "unlocked-io.h" be missing. So you need your own substitute > unlocked-io.h, which (presumably) does nothing.
Yes, right. The user of --avoid=unlocked-io would have to provide an empty unlocked-io.h file. > > Regarding xalloc-die, I propose this patch. > > This doesn't look quite right to me. xalloc-die depends on error, > gettext-h, and exitfail, because its implementation uses those three > modules. But a user who substitutes a specially-built xalloc-die > implementation may not need those three modules. It's common to take > the transitive closure of the dependencies to see which modules your > app needs. In that scenario, user would end up with modules like > "error" that they don't need. The way it's implemented in gnulib-tool, the --avoid acts _during_ the computation of the transitive closure, not afterwards. Which means, --avoid=xalloc-die prevents the xalloc-die module from being included AND also prevents the error, gettext-h, exitfail modules from being included if nothing else needs them. Bruno _______________________________________________ bug-gnulib mailing list bug-gnulib@gnu.org http://lists.gnu.org/mailman/listinfo/bug-gnulib