On Thu, Mar 29, 2012 at 02:23:02PM +0200, David Sterba wrote:
On Thu, Mar 29, 2012 at 08:06:25AM -0400, Darryl L. Pierce wrote:
I'm experiencing the following error (have been for a while which is why I've been stalled):
gcc -DPKGDATADIR="/usr/local/share/mutt" -DSYSCONFDIR="/usr/local/etc" -DBINDIR="/usr/local/bin" -DMUTTLOCALEDIR="/usr/local/share/locale" -DHAVE_CONFIG_H=1 -I. -I. -I. -Iintl -I./intl -Wall -pedantic -Wno-long-long -g -O2 -MT buffy.o -MD -MP -MF .deps/buffy.Tpo -c -o buffy.o buffy.c buffy.c: In function ‘mutt_buffy_check’: buffy.c:602:21: error: ‘VirtIncoming’ undeclared (first use in this function) buffy.c:602:21: note: each undeclared identifier is reported only once for each function it appears in make[2]: *** [buffy.o] Error 1 make[2]: Leaving directory `/home/mcpierce/Programming/mutt-kz' make[1]: *** [all-recursive] Error 1 make[1]: Leaving directory `/home/mcpierce/Programming/mutt-kz' make: *** [all] Error 2
You need to run configure with --enable-notmuch. Somewhere it's used out of the #ifdef NOTMUCH/#endif block.
Since part of our goal (I assumed) was integration of notmuch with the Mutt code, why do we not have this by default? It would seem to make more sense for regression tests to have to explicitly disable rather than enable notmuch.