Hi there. I'm porting a system from Centos 5 to Centos 6, and have observed a difference in m4 behavior. I'd like to know if this change was intentional and if there is a way (short of installing the older version) to get the old behavior. I've read through the changelog entries, but haven't found anything.
m4 is being used to turn a sendmail.mc file in to a sendmail.cf file. The OSTYPE line refers to "linux-gnu" but there is no /usr/share/sendmail-cf/ostype/linux-gnu.m4 file in this case, which generates a "no such file or directory" warning. In version 1.4.5 (Centos 5) this causes the same warning, but m4 still returns a return code of 0 and successfully completes. In version 1.4.13 (Centos 6) m4 returns a return code of 1, even though it still can generate the file. It *appears* like the -E option has become an undocumented default behavior, turning warnings into, effectively, errors. Is this correct? Is there a way to revert this behavior? In my particular case, I could use the "linux.m4" or "gnu.m4" files present on the system or create a new linux-gnu.m4 file as a work-around, however that would only fix this one case. The code could be installed on other systems with different missing m4 files, and I'd like to make *any* missing "ostype/*file.m4" be a non-error-code-setting warning again, as it was in version 1.4.5. Any insight into how/why this change happened, and a way to work around it to get the old behavior would be greatly appreciated. Thanks, Jesse
