13/03/2019 12:20, Bruce Richardson: > On Wed, Mar 13, 2019 at 12:13:52PM +0100, Thomas Monjalon wrote: > > 07/03/2019 14:35, Bruce Richardson: > > > +# > > > +# Version information completed when this file is processed for a build > > > +# > > > +CONFIG_RTE_VER_YEAR=__YEAR > > > +CONFIG_RTE_VER_MONTH=__MONTH > > > +CONFIG_RTE_VER_MINOR=__MINOR > > > +CONFIG_RTE_VER_SUFFIX=__SUFFIX > > > +CONFIG_RTE_VER_RELEASE=__RELEASE > > [...] > > > -/** > > > - * Patch release number > > > - * 0-15 = release candidates > > > - * 16 = release > > > - */ > > > -#define RTE_VER_RELEASE 0 > > > > So you are removing the special value 16. > > Is it replaced by an empty value? > > > > I think it is an issue for version comparison with RTE_VERSION_NUM() macro. > > The -rc3 must be lower than the final release (which had number 16 > > for this reason). > > > Aha, that explains the reason for the 16 value. I did wonder why we > bothered with it when it wasn't actually used in printing etc. Let me go > back and look into this set again, armed with this new info. [We also need > to document this reason in the code comments for future information]
Yes it was not documented, sorry. Thanks for working on it.