Ryo ONODERA <ryo_on%yk.rim.or.jp@localhost> writes: > Is there any reason to depend on curses-based pinentry only? > For example, in thunderbird enigmail case, > pinentry-gtk2 works however curses-based pinentry does not work for me. > MESSAGE or other documentation file should suggest pinentry* packages > installation. In general, I view the use of MESSAGES as a bug. It was originally for exceptional situations where the machine admin had to do something, and it's grown into a substitute for documentation. > I know gnupg2 package has the dependency to pinentry. > i believe gnupg2 should not have pinentry dependency. I can see your point, in that gnupg2 is usable without it, and if one has pinentry-gtk2, there may be no need for many situations. However, the real reason to avoid gnupg2 depending on pinentry-gtk2 is that it requires gtk2, which is big. The pinentry package is 136K (netbsd-6 i386), and it only depends on things already required by gnupg2. So I don't think it hurts enough to remove it. If pinentry-gtk2 and pinentry-qt4 could be added for 150K each of footprint, I'd be in favor. But pinentry-gtk2 and dependencies is 65MB, and pinentry-qt4 is 98 MB. All that said, because thunderbird is a gtk thing, not a qt thing, having the enigmail plugin depend on pinentry-gtk2 seems reasonable. Sort of realted, are we at the point where the gnupg2 package should build gpg, and gpg 1 should be deprecated? I'm not clear on why we are still using gpg1.
Attachment:
pgpeTbA5I3Af0.pgp
Description: PGP signature