[uClibc]Rules.make $(CROSS) change

Erik Andersen andersen at codepoet.org
Fri Nov 8 05:09:06 UTC 2002


On Fri Nov 08, 2002 at 01:28:29PM +0900, Miles Bader wrote:
> Is this change to `Rules.make' actually necessary?  Why was it done?
[---------snip-------]
>    +CC:= $(CROSS)gcc
>    +AR:= $(CROSS)ar
>    +LD:= $(CROSS)ld
>    +NM:= $(CROSS)nm
>    +STRIPTOOL:= $(CROSS)strip

I can revert the ":=" to just "=" if you want.  Seemed like it 
should speed up compiling (one less thing to evaluate 10 million
times).

> As I mentioned in an earlier message, it's something of a regression
> that you have to specify _anything_ on the make command line, and having
> to specify CROSS is especially ugly -- it's not a `nice' value (it feels
> like an internal implementation artifact).

I agree.  But I don't really have a good way to work around it.
The linux kernel requires you specify CROSS (either by editing
the top level Makefile, or by specifying it on the command line).
We effective now require the same sortof thing.

I agree with you, that was one nice thing about the old config
system that we have lost in the transition.  Of course, with the
old config system you had to copy or link the correct Config file
into place, so it wan't totally free.  On the plus side, we have
gained 'make oldconfig', which is a much nicer way to update a
config file then the old system...

> I worked around this need to specify CROSS on the v850 by adding a hack
> that adds CROSS as a config variable; this makes unecessary to specify
> any command-line arguments at all during normal makes (though you still
> have to specify TARGET_ARCH during config makes [I think that should be
> fixed as well]).  The above changes breaks this.

K, I'll revert to just "=".

BTW, I'm trying to get a release out tonight (more testing on the
new config system).  So if you get a chance to get a
extra/Configs/Config.v850.default file into CVS, that would be
cool.

 -Erik

--
Erik B. Andersen             http://codepoet-consulting.com/
--This message was written using 73% post-consumer electrons--



More information about the uClibc mailing list