RE: .SECONDEXPAND and spaces?

[Paul D. Smith]

%% "Leeuwesteijn,Joost" <address@hidden> writes:

  l> Is there a big performance penalty when putting .SECONDEXPANSION at
  l> the top of my (toplevel) makefile? Probably not in my small
  l> project, but in general.  Could it be a problem in large project? I
  l> suppose not because the files will be loaded to memory only once;
  l> it's all done in memory which shouldn't really make a big
  l> difference.

There is a very slight performance penalty in that when .SECONDEXPANSION
is enabled make performs a second expansion pass in various places where
it doesn't need to otherwise.

  l> Could .SECONDEXPANSION have any other unwanted side-effects I
  l> should worry about?

If you have "normal" prerequisite files that contain "$" in the names
(many Java files do this for example) then you'll have big problems
enabling .SECONDEXPANSION.

Otherwise, there is no issue.

  >> -----Original Message-----
  >> 
  >> When I try to do secondary expansion, it seems to work. 
  >> Adding a simple string to the target name:
  >> $(MOD1.OBJECTS) : address@hidden
  >> gives a make -p output of:
  >> OUTPUT/MOD1/mod1file2.o: OUTPUT/MOD1/mod1file2.o-fubar 
  >> Which is OK.
  >> 
  >> 
  >> When I try to mangle the prerequisite a bit more (just a  test)
  >> it doesn't seem to work:
  >> $(MOD1.OBJECTS) : $(MOD1.SOURCEDIR)/$$(notdir $$@)
  >> gives a make -p output of:
  >> OUTPUT/MOD1/mod1file1.o: MOD1/$(notdir) MOD1/$($@)
  >> Which is not what I expected.

It sounds like you're using an older version of GNU make.

The .SECONDEXPANSION feature is only available starting in GNU make
3.81.

-- 
-------------------------------------------------------------------------------
 Paul D. Smith <address@hidden>          Find some GNU make tips at:
 http://www.gnu.org                      http://make.paulandlesley.org
 "Please remain calm...I may be mad, but I am a professional." --Mad Scientist