On Thu, Aug 03, 2023 at 11:18:09AM +0200, David Marchand wrote: > On Mon, Jul 17, 2023 at 1:09 PM Bruce Richardson > <bruce.richard...@intel.com> wrote: > > > There may be some polishing to do later. > > > Looking at the result for rte_eal_init, I see that the generated > > > manual starts with a reference to the rte_eal.h header with a path > > > relative to its location in the DPDK tree. > > > $ MANPATH=build-gcc/install/share/man man rte_eal_init | head -5 > > > lib/eal/include/rte_eal.h(3) > > > DPDK > > > > > > lib/eal/include/rte_eal.h(3) > > > > > > NAME > > > lib/eal/include/rte_eal.h > > > > > > At least, it is possible to ask for this header man with "man > > > rte_eal.h", but it is a bit confusing. > > > Is there something we can do on this side? > > > > > > > Not sure, not really familiar with how doxygen works generating manpages > > and the options supported, etc. etc. Mainly I just looked at the > > build-system side to support this, since I really missed having manpages > > for DPDK functions to quickly check parameter order. > > doxygen is invoked on the sources tree, so this is probably the reason > why such path is in the generated manual. > Maybe Thomas has an idea how we could handle this. > > In any case, I am fine with this patch as it is now. > I think I've found a fix for this. doxygen setting "FULL_PATH_NAMES" defaults to "YES", which is what we want for the HTML pages. However, setting it to "NO", I believe is correct for generating the manpages.
Patch V3 on its way. /Bruce
