I agree with backing out the change. Not for the fact that it causes
Configure to exit, but for the fact that it leaves Configure no control
over whether it exits. Better to return a value reporting lack of
success, that Configure can use to exit if it chooses to do so. (I'll
say more in reply to other James' other message.)
Allison
James E Keenan wrote:
Paul,
I have to object to this recent modification to config/init/manifest.pm:
r21314 | paultcochrane | 2007-09-17 03:04:46 -0400 (Mon, 17 Sep 2007) |
6 lines
[config] If some files are missing in the repository, but are mentioned
in the MANIFEST then Configure complains, saying that it can't continue
without these files. However, Configure continues anyway, which is not
the intended behaviour. This commit now causes Configure to exit
instead of just return at this point.
Index: config/init/manifest.pm
===================================================================
--- config/init/manifest.pm (revision 21271)
+++ config/init/manifest.pm (working copy)
@@ -45,7 +45,7 @@
END
- return;
+ exit 1;
}
return $self;
I object to this because it implicitly short-circuits a discussion we've
been having in another thread about what Configure.pl should do when a
particular configuration step fails.
If, for example, we haven't yet collectively come to the conclusion that
configuration should stop if it can't find a valid C-compiler, then
we're not at a point where we can collectively say that configuration
should stop if a particular file is missing from the MANIFEST. As I
have argued recently, we need to place the discussion of how to handle
the failings of a particular configuration step in the context of an
overall discussion.
(As a matter of fact, config/init/manifest.pm is one of the few config
steps that, IMO, handles failure well.)
I recommend that you back this change out of head and file an RT
proposing the change.
kid51
