On Tuesday, March 01, 2011 02:20:19 Thomas Petazzoni wrote: > This workaround sounds a lot uglier than the patch I propose, which > cleanly makes docbook2man non-essential. not really. looking at the patch again: +if HAVE_DOCBOOKTOMAN +MANPAGES = $(MAN5) $(MAN8) +endif +dist_man_MANS = $(MANPAGES) the man pages wont even be installed now unless docbook is installed. that's even worse than the issue i raised earlier. > Would you accept a patch that adds a --enable-doc/--disable-doc option. > --enable-doc is the default value, and errors out at the ./configure > step if docbook2man is not available. --disable-doc would not require > docbook2man and would not generate the documentation. > > This way, the release process is unchanged: the one that > runs ./configure by default is required to have docbook2man, but all > those embedded Linux build systems can pass --disable-doc if they don't > care about the documentation. you're basically trying to hack around a broken release process. i think it makes more sense to fix the release process and then your patch isnt necessary. your proposal here is better than the original one (and ultimately, i'm not the m-i-t upstream maintainer), but you still need to address the new bug i raised above. -mike
Attachment:
signature.asc
Description: This is a digitally signed message part.
