Re: inconsistency of 'emptiness'?

[gk]

Hi Paul,
Thanks for your input.

At 04:29 PM 1/28/2003 -0500, Paul D. Smith wrote:
> Can you please summarize your ultimate findings into a bug report and
> submit it on the Savannah site for me?  That way I'll be able to
> remember it.  I'll see if I can determine why make is not handling the
> backslash consistently.

Sure. Will do.


> I'm leaning towards not fixing the initial issue you reported, where:
>
>   $ make FOO=' '
>
> results in FOO being empty rather than containing a single space,
> because to fix that would mean that the behavior of this:
>
>   $ make 'FOO  =  bar'
>
> which currently works "as you'd expect" from a make variable setting and
> gives a value of "bar" instead of " bar", would change incompatibly.


That sounds reasonable to me.


> But, I do wonder about the backslash issue.

Yeah. That is the only really weird thing.
It seems like a reasonable thing would probably just be for make to split 
the parameter 'FOO=   value or \ $$whatever' on the '=' and assign the 
makefile variable FOO to everything on the right of '=', IN THE MAKEFILE 
CONTEXT - exactly as what would happen in a makefile if the exact same 
string value were assigned: i.e., leading space gets lost and any backslash 
in value is assigned to FOO, any makefile variables like $$whatever get 
expanded, as happens now.  I think that would be very consistent.


- Greg Keraunen
http://www.xmake.org
http://www.xmlmake.com