Re: [Flex-devel] FW: Configure CPP With ARM Hard-Float

[Will E. <wes...@gm...>]

On Tuesday, 29 April 2014,  7:29 pm -0500, Nathan Royce <na...@ho...> wrote:

> I don't know why it was rejected. I didn't think it was appropriate for "flex-help" since I wasn't using it yet for any other applications.

You're likely not a member of the list.

> 
> 
> Subject: Configure CPP With ARM Hard-Float
> From: fle...@li...
> To: na...@ho...
> Date: Tue, 29 Apr 2014 14:17:34 +0000
> 
> You are not allowed to post to this mailing list, and your message has
> been automatically rejected.  If you think that your messages are
> being rejected in error, contact the mailing list owner at
> fle...@li....
>  
> 
> 
> --Forwarded Message Attachment--
> From: na...@ho...
> To: fle...@li...
> Subject: Configure CPP With ARM Hard-Float
> Date: Tue, 29 Apr 2014 09:17:26 -0500
> 
> 
> 
> 
> I was having an issue with configuring flex and found that CPP was being set to "/lib/cpp". Looking at the line 7150 "for CPP in "$CC -E" "$CC -E -traditional-cpp" "/lib/cpp"", I see the test does not take into account any flags and the particular one in CFLAGS is "mfloat-abi=hard", otherwise it looks for a missing "stubs-soft.h"
> I had a "make check" problem with lzlib and plzip with the same issue and submitted to that as well, and is fixed in the development releases. The neat trick the developer told me before it was fixed was to set "LDFLAGS='$CFLAGS $LDFLAGS'" which I used in this flex case by setting CC='$CC $CFLAGS'.I have a feeling I'm going to encounter this with many packages. 		 	   		   		 	   		  

Thanks for your report. I'm rewriting the test suite for a number of reasons, as it happens, so I'll watch out for this sort of thing. $CPPFLAGS, $CFLAGS and $LDFLAGS are all different variables that should be set independently.