On Sat, Dec 24, 2022 at 02:16:20PM -0500, Steven Rostedt wrote: > On Sat, 24 Dec 2022 16:32:25 +0100 > Daniel Wagner <dwagner@xxxxxxx> wrote: > > > During my packaging attempt I found out that the man pages are installed > > into subdirs. So I started to look into the this and ... let's say it's > > complicated. > > Oh, to make your life easier. That shouldn't happen. To be honest most of the Meson stuff was pretty easy to get sorted out. Generating documentation seems to be a good source of troubles. I had also headache to sort this out for libnvme/nvme-cli. > > For example libtraceevent-func_apis.txt list a few function which are > > also listed in libtraceevent-func_find.txt. The Makefile uses these > > function as build names but, if we have two targets with the same name > > one will overwrite the other. make seems to tolerate something like > > this. Meson doesn't. > > That's actually a mistake :-/ > > The asciidoc complains whenever a man page has more than 9 functions > described, so I have to break them up when I add more related APIs. > > I probably missed removing some of the old functions when I did that > here. Ah, that explains it. > > I am playing with various ideas how we could handle this. Currently, I > > think the best way is to distinguish the different source txt files. So > > we have files which are for generating a mans per function (1:n) > > and we have txt files which are general documentation (1:1). > > The real fix is to remove the tep_find_function() and > tep_find_function_address() from libtraceevent-func_apis.txt. Right, let me add this fix then. Anyway, I got a new version almost ready. I try to package it with the current version, verifying I got it sorted out.
|