On Tue, Mar 14, 2017 at 4:34 AM, Mathieu Lirzin <m...@gnu.org> wrote: > Jim Meyering <j...@meyering.net> writes: > >> On Tue, Mar 7, 2017 at 8:08 AM, Mathieu Lirzin <m...@gnu.org> wrote: >>> >>> Jim Meyering <j...@meyering.net> writes: >>> >>>> On Sun, Mar 5, 2017 at 8:12 AM, Mathieu Lirzin <m...@gnu.org> wrote: >>>>> >>>>> I have been using 'readme-release' module and was surprised to discover >>>>> that the "README-release" file was automatically distributed without any >>>>> mention in the "module/readme-release" file. >>>>> >>>>> After some digging I have discovered that all files from the "top" >>>>> directory are automatically added to EXTRA_DIST. Here is a snippet from >>>>> bootstrap that could be removed: >>>>> >>>>> --8<---------------cut here---------------start------------->8--- >>>>> if test -n "$top_files"; then >>>>> sed_prepend_topdir='s,^,$(top_srcdir)/,' >>>>> echo "EXTRA_DIST += "`echo "$top_files" | sed -e >>>>> "$sed_prepend_topdir"` >>>>> echo >>>>> fi >>>>> --8<---------------cut here---------------end--------------->8--- >>>>> >>>>> IMO we should move the responsability of distributing "top/*" files to >>>>> the module definition instead, and not distributing "README-release" >>>>> automatically (like what is done in Coreutils). >>>>> >>>>> What do people think? >>>> >>>> Regarding README-release, I'm leery of requiring each >>>> top-file-specifying module writer to remember to add each such file to >>>> EXTRA_DIST. An alternative, probably-smaller change may be to add some >>>> attribute by which a module with such a file could opt out of that >>>> default. >>> >>> That would be fine with me. However I am not sure this additional logic >>> would be smaller (or simpler), since IIUC there is currently only 3 >>> top-file-specifying modules: >>> >>> - maintainer-makefile >>> - readme-release >>> - gnumakefile >>> >>> Do I overlook something? >> >> I hadn't counted, and expected more. Given there are only three, >> adding an 'EXTRA_DIST += ...' in two of the three seems reasonable. Do >> you feel like writing the patch? > > Here is the patch:
At first glance, that diff looks perfect. The only missing piece (actually important, since not easily derived by looking at the patch) is justification: why make this change. Would you please adjust the log/ChangeLog to say it is to stop distributing README-release?
