over in https://bugs.debian.org/833481, Adam Borowski notes that --yes often
does not work unless --batch is also supplied.

Looking at the source, i see that --yes is mapped to oAnswerYes, which is in
turn mapped to opt.answer_yes.

0 dkg@alice:~/src/gnupg$ git grep answer_yes
g10/ChangeLog-2011: cpr_get_answer_is_yes,cpr_get_answer_yes_no_quit): Modified
to work
g10/ChangeLog-2011: * status.c (cpr_get_answer_yes_no_quit): New.
g10/cpr.c:cpr_get_answer_yes_no_quit( const char *keyword, const char *prompt )
g10/delkey.c: else if (opt.batch && opt.answer_yes)
g10/delkey.c: opt.answer_yes);
g10/gpg.c: case oAnswerYes: opt.answer_yes = 1; break;
g10/keyedit.c: if (opt.batch && opt.answer_yes)
g10/keyedit.c: okay = cpr_get_answer_yes_no_quit
g10/keyedit.c: okay = cpr_get_answer_yes_no_quit
g10/keyedit.c: okay = cpr_get_answer_yes_no_quit
g10/keygen.c: use_tty = (!opt.batch && !opt.answer_yes
g10/keygen.c: if (opt.answer_yes)
g10/main.h:int cpr_get_answer_yes_no_quit( const char *keyword, const char
*prompt );
g10/openfile.c: if (opt.answer_yes)
g10/options.h: int answer_yes; /* answer yes on most questions */
g10/photoid.c: switch(cpr_get_answer_yes_no_quit("photoid.jpeg.okay",
g10/trustdb.c: if (opt.batch && !opt.answer_yes)
sm/gpgsm.c: case oAnswerYes: opt.answer_yes = 1; break;
sm/gpgsm.h: int answer_yes; /* assume yes on most questions */
0 dkg@alice:~/src/gnupg$

It appears two places don't require opt.batch:

• openfile.c
• keygen.c

and three do:

• trustdb.c
• delkey.c
• keyedit.c

If there's a pattern to this, it's not clear what it is. Could the pattern be
made simpler (e.g. --yes only works with --batch everywhere; or --yes works
regardless of --batch)? And could the behavior be explained clearly in the
usual place where people look for documentation (e.g. doc/gpg.texi)?