Re: typo in the make manual

[Philip Guenther]

On Sun, Apr 17, 2011 at 5:29 PM, Paul Smith <address@hidden> wrote:
> On Tue, 2011-04-05 at 07:40 +0430, ali hagigat wrote:
>> 6.12 Pattern-specific Variable Values
>> If a target matches more than one pattern, the matching
>> pattern-specific variables with
>> longer stems are interpreted first.
>> -------------------------------------------
>> The above words of the documentation seems to be corrected with
>> "shorter stems".
>
> No, the manual is correct.  The longer stems are preferred over shorter
> stems.

Paul, I'm not sure "preferred" is the right word there.

The patterns-specific variable assignments are applied starting from
the most generic pattern (i.e., the longest stems) to the
most-specific pattern (i.e., the shortest stem).  For normal
assignment (=, :=), that will result in the assignment from the
most-specific pattern having precedence, as if the more generic
assignment wasn't there.  However, for += and ?= assignments, you can
'detect' the assignment from the more generic pattern.  Consider:

$ cat Makefile
%: type1 = generic
%: type2 = generic
%: type3 ?= generic
%.c: type1 = C
%.c: type2 += C
%.c: type3 ?= C

foo.c:
        echo '${type1}'
        echo '${type2}'
        echo '${type3}'
$ make
echo 'C'
C
echo 'generic C'
generic C
echo 'generic'
generic
$

For ${type1}, the %.c assignment overrode the % assignment.
For ${type2}, the %.c assignment added onto the % assignment.
For ${type3}, the %.c assignment had no effect, because type3 was
already set by the % assignment.


So the manual is correct: longer (more generic) stems are interpreted
before shorter (more specific) stems...but you'll probably be using
normal assignment so that the latter will override the former.


Philip Guenther