Re: .PHONY and .SECONDARY interaction in make 3.81: bug/feature?

[Alexey Neyman]

A little more info:

revision 1.69 made .SECONDARY target not only set the 'all_secondary' 
variable to 1, but also sets ->intermediate flag on all files. 
Therefore, the check_dep() function in remake.c doesn't notice that 
version.c needs to be rebuilt unconditionally.

The following patch solves the problem. No regressions introduced on 
i686-pc-linux-gnu (Fedora Core 2).

*** remake.c.orig       2005-03-15 19:07:49.213951064 +0300
--- remake.c    2005-03-15 19:01:25.309097872 +0300
*************** check_dep (struct file *file, unsigned i
*** 908,914 ****
    ++depth;
    start_updating (file);

!   if (!file->intermediate)
      /* If this is a non-intermediate file, update it and record
         whether it is newer than THIS_MTIME.  */
      {
--- 908,914 ----
    ++depth;
    start_updating (file);

!   if (!file->intermediate || file->phony)
      /* If this is a non-intermediate file, update it and record
         whether it is newer than THIS_MTIME.  */
      {

Regards,
Alexey.

On Tuesday 15 March 2005 18:50, Alexey Neyman wrote:
> I tried make 3.81beta2, and its behavior differed from make 3.80 for 
> some of my Makefiles I used. I tracked the problem to the following 
> testcase:
> 
> ======= Makefile ========
> OBJS            = main.o version.o
> all: main
> clean:
>         rm -f main $(OBJS) version.c
> 
> date_stamp      := $(shell LANG=C date)
> 
> .SECONDARY:
> .PHONY: version.c
> version.c:
>         @echo "Creating address@hidden"
>         @echo "const char *version=\"$(date_stamp)\";" > $@
> 
> %.o: %.c
>         gcc -c -o $@ $<
> 
> main: $(OBJS)
>         gcc -o $@ $^
> =========================
> 
> The 1st "make all" does what's expected with both make 3.80 and make 
> 3.81: rebuilds the 'main' binary. However, if "make all" is issued 
> again, make 3.80 rebuilds version.c and all files that depend on it 
> (version.o, main). Make 3.81beta2, however, considers it up-to-date 
> and doesn't rebuild anything.
> 
> I didn't find anything in the description of the .SECONDARY that 
> states such behavior.
> 
> This behavior was introduced by the revision 1.69 of the 'file.c' 
> file. The description says that it fixes bug #2515, but that bug 
ends 
> with a sentence "I'll look at this more carefully." by Paul Smith.
> I am not sure which .SECONDARY behavior is right.
> 
> PS. I know that .SECONDARY is pointless in this makefile, this 
> testcase was snipped from a much larger project.
> 
> Regards,
> Alexey.
> 
> -- 
> We seek not to invade, but to pervade.  There is a difference.
>                         -- Arilou, SC2
> 
> 
> _______________________________________________
> Help-make mailing list
> address@hidden
> http://lists.gnu.org/mailman/listinfo/help-make
> 

-- 
Do not ask favors of people you routinely ambush and murder.
                        -- Spathi captain Fwiffo, SC2