On Fri, Dec 20, 2024 at 01:50:25PM -0800, Junio C Hamano wrote: > Patrick Steinhardt <ps@xxxxxx> writes: > > > diff --git a/Documentation/Makefile b/Documentation/Makefile > > index ff30ab6c4295525757f6a150ec4ff0c72487f440..a89823e1d1ee5042367bdcca6ed426196d49ce89 100644 > > --- a/Documentation/Makefile > > +++ b/Documentation/Makefile > > @@ -181,6 +181,10 @@ endif > > -include ../config.mak.autogen > > -include ../config.mak > > > > +# Set GIT_VERSION_OVERRIDE such that version_gen knows to substitute > > +# GIT_VERSION in case it was set by the user. > > +GIT_VERSION_OVERRIDE := $(GIT_VERSION) > > + > > ifndef NO_MAN_BOLD_LITERAL > > XMLTO_EXTRA += -m manpage-bold-literal.xsl > > endif > > So the idea is that those targets and scripts may have their own > GIT_VERION value when they run GIT-VERSION-GEN to cause GIT_VERSION > to computed, and in such a case, they should pass the GIT_VERSION > they have in GIT_VERSION_OVERRIDE, and thanks to the version_gen > thing, this value in GIT_VERSION_OVERRIDE is passed in the > environment as GIT_VERSION when GIT-VERSION-GEN is run, and the > value in turn is passed intact. Somehow this makes my head spin, as > it looks quite convoluted, but the overall flow should yield the > desired value. Agreed, it makes mine spin, as well. Next release cycle I'll have a look at whether I can get rid of the include of "GIT-VERSION-FILE" altogether to make the logic simpler. Patrick