--disable-gtk-doc is parsed by the gnome2.eclass: # Preserve old behavior for older EAPI. if grep -q "enable-gtk-doc" "${ECONF_SOURCE:-.}"/configure ; then if has ${EAPI:-0} 4 && in_iuse doc ; then g2conf+=( $(use_enable doc gtk-doc) ) else g2conf+=( --disable-gtk-doc ) fi fi but this is now controlled by the gtk-doc useflag, so the same argument can be parsed twice: --disable-gtk-doc --disable-maintainer-mode --disable-schemas-compile --disable-quartz-backend --disable-broadway-backend --disable-cloudprint --disable-colord --enable-cups=auto --disable-gtk-doc or contradictory: --disable-gtk-doc --disable-maintainer-mode --disable-schemas-compile --disable-quartz-backend --disable-broadway-backend --disable-cloudprint --disable-colord --enable-cups=auto --enable-gtk-doc
Sure, all that is correct, but what is your problem with it?
So it's not a problem when one compile argement is parsed twice, or two contradictory ones are passed?
No, it's not a problem. With configure the last one of the same kind wins -- the --enable-gtk-doc will override the earlier --disable-gtk-doc from eclass. Note that USE=gtk-doc on gtk+ merely regenerates the gtk-docs - you get them even without USE=gtk-doc. I added them because the pregenerated ones in tarballs lacked "Index of new symbols since 3.24" and co index sections, which I needed for my own work and thought them to be useful to be available for others that set global gtk-doc too (which is more used now with meson, as there you don't have any pregenerated docs installed without a USE flag)