I encountered a surprising corner case recently; not clear whether it constitutes a bug.
The documentation is clear that when an include file is not found make sits on the message for a while in case the file can be made. That I understand, but there are two surprising behaviors which result. First, only the most recent missing include file message seems to be cached: $ cat Makefile include no_such_file_1 include no_such_file_2 include no_such_file_3 all:;@: $ make Makefile:3: no_such_file_3: No such file or directory make: *** No rule to make target 'no_such_file_3'. Stop. Second, if make is aborted by $(error ...) no cached messages are printed at all: $ cat Makefile include no_such_file_1 include no_such_file_2 include no_such_file_3 $(error DIE HERE) all:;@: $ make Makefile:4: *** DIE HERE. Stop. Imagine a config file which is supposed to set BASEDIR followed by an assert to make sure it's present: include cfgfile $(if $(BASEDIR),,$(error Error: BASEDIR not set)) The result of not finding cfgfile is *only* the "BASEDIR not set" error - nothing about the include file. It's easy to see how a person could waste lots of time going through the code in cfgfile trying to figure out why it doesn't set BASEDIR before eventually realizing it isn't included at all. So: bug or just-how-it-is? David
_______________________________________________ Bug-make mailing list Bug-make@gnu.org https://lists.gnu.org/mailman/listinfo/bug-make
