Make deletes intermediate files, even though I use .SECONDARY or .PRECIOUS - makefile

At the end of my build, make deletes a file:
Removing intermediate files...
rm some/dir/myfile.inc
I want to keep it to make later builds faster, but I have not been able to. Either one of these two lines should force make to keep all intermediate files, as in this question, but they have no effect:
.SECONDARY:
.PRECIOUS:
If I spell out the exact file path it does keep the file:
.SECONDARY: some/dir/myfile.inc
If but this does not work
.SECONDARY: %.inc
Is the .SECONDARY rules being ignored because of a later rule like this is more specific?
%foo.o %foo.h %foo.inc: some_other_file
dosomething

Related

Is it ok to have a GNU Make target that claims to generate / update a certain target file but actually doesn't?

At present, I have a makefile that has:
a target which links an executable image file from a bunch of object files
a pattern rule target that compiles the various object files the linker target depends on
I want to make the following changes.
Instead of compiling the object files outright, I want the pattern rule target mentioned above to create (for each object file that needs updating) an empty object_file_name.update file. Essentially, this target's job would be to take stock of all object files that actually need to be recompiled.
Write a new target that launches a Perl process which finds all these object_file_name.update files and, for each object file that must be recompiled, compiles it in this Perl process.
I know how to do 2) ... that part is not giving me any trouble. The part I'm worried about is 1). The reason is that that target would basically have to claim to update any needed object files while, in truth, merely creating an .update file for each such object file but not the object file itself.
I think I could trick GNU Make into not starting to try to link anything before all the object files have been built by declaring my dependencies accordingly (pseudo-code, not a valid GNU Make snippet):
# Phony target that reads the *.update files created by the pattern rule target below and then
# compiles each object file for which an *.update file exists.
COMPILE_OBJECTS :
...
# Pattern rule target to take stock of all object files that need updating. Creates an *.update file for
# each object file that needs recompiling.
%.o : %.c :
...
$(EXE_FILE_TO_LINK) : $(LIST_OF_OBJECT_FILE_PATHS) COMPILE_OBJECTS
...
but I still worry that this might result in undefined behavior because my pattern rule target would basically be lying to GNU Make about updating the needed object files. Is my worry justified?
Basically, I want to interject an intermediate layer between GNU Make and the compiler so that GNU Make doesn't compile each object file separately. Instead, the compiling would be done in a single Perl process that has access to the complete list of object files that need to be compiled, allowing me to do various fancy things that I couldn't do if GNU Make controlled compilation directly.
Yes, it's legal and I often use this pattern.
Consider the case where you only want to kick off a long build step if a file has changed.
target: config-file
target-creator $< -o $#
Now let's say we can't give make the dependencies for config-file (because the config file creation step lacks a dependency listing ability (BAH!)).
.PHONY: FORCE
FORCE: ;
config-file: FORCE
config-creator -o $#.tmp
cmp $#.tmp $# || mv $#.tmp $#
We ask make to build target
Make first has to build config-file
Make will always run the recipe for config-file,
as its dependency FORCE is out of date (being phony)
CRUCIALLY we only update config-file if config-creator decides something has actually changed
If cmp decides config-file.tmp and config-file are the same, and the last line of the recipe completes with no error
OTOH if cmp detects a mis-compare, it fails, and the shell goes on to execute the mv.
After running the recipe for config-file, make does actually check config-file's modification time. IF config-file has become younger than target, only then will target-creator be run.
The subtlety here is that even though config-file's recipe runs every time, config-file itself is not phony.

How to trigger the rebuild of a Makefile prerequisite file ONLY when a specific target is called?

I haven't found an answer so far, so I think this is not a repeat question.
I have some Makefile along the lines of:
include prerequisite_2
all: prerequisite_1 prerequisite_2
clean:
rm *.mod
prerequisite_1:
mkdir somedir
prerequisite_2:
re-write existing file
The issue is that I want the prerequisite_2 to rebuild whenever the default goal is called (all) or when prerequisite_2 is called from the command line, and I know I can use touch prerequisite_2, FORCE or .PHONY to achieve this. However, I DO NOT want it to run every time (the written file contains dependency information for the Fortran files involved) as it doesn't make sense to also rebuild this when calling: make clean
Is it possible to emulate the effects of FORCE or .PHONY only when the depending targets are called?
You can see what the goal targets are by looking at the MAKECMDGOALS variable.
So you can do something like:
ifeq (,$(if $(MAKECMDGOALS),$(filter-out all prerequisite-2,$(MAKECMDGOALS))))
include prerequisite-2
endif
The if condition will be true if MAKECMDGOALS is the empty string, or if it contains only all and/or prerequisite-2 but not if it contains any other target.
Usually, this is not what you want though. Usually you want to disable the include only if certain targets (clean is the classic example) are used.
This exact situation is even discussed in the GNU make manual.

Makefile rule only works if file exists before make is invoked

Consider the following (MCVE of a) Makefile:
my_target: prepare test.bin
prepare:
echo >test.dat
%.bin: %.dat
cp $? $#
If you run make in a clean directory, it fails:
echo >test.dat
make: *** No rule to make target 'test.bin', needed by 'my_target'. Stop.
Run it again and it succeeds:
echo >test.dat
cp test.dat test.bin
What seems to happen is that the rule to make *.bin from *.dat only recognises that it knows how to make test.bin if test.dat exists before anything is executed, even though according to the output it has already created test.dat before it tries to create test.bin.
This is inconvenient for me as I have to prepare a few files first (import them from a different, earlier part of the build).
Is there a solution? Perhaps some way to allow the rules to be (re)evaluated in the light of the files which are now present?
There are a number of issues with your makefile. However based on your comments I'm inclined to assume that the MCVE here is just a little too "M" and it's been reduced so much that it has a number of basic problems. So I won't discuss them, unless you want me to.
The issue here is that you're creating important files without indicating to make that that's what you're doing. Make keeps internally a cache of the contents of directories that it's worked with, for performance reasons, and that cache is only updated when make invokes a rule that it understands will modify it.
Here your target is prepare but the recipe actually creates a completely different file, test.dat. So, make doesn't modify its internal cache of the directory contents and when it checks the cache to see if the file test.dat exists, it doesn't.
You need to be sure that your makefile is written such that it doesn't trick make: if a recipe creates a file foo then the target name should be foo, not bar.
This happens for wildcard targets, like %.bin. They get evaluated at the first pass. You could add an explicit target of test.bin. Or, follow the advice of tkausl and have test.dat depend on prepare (a phony target). In this case, you don't need the double dependency anymore:
my_target: test.bin
you have to write
test.dat: prepare
or (when when you want to stay with wildcards)
%.dat: prepare
#:
Usually, you might want to create and use .stamp files instead of a prepare target.

Makefile how to not remove one specific object file while remove all other object files and executable files?

Right now I have this command: rm -f *.x *.o, which removes all the object files and executable files by calling make clean in my shell. But I have one specific object file, say a.o that I do not want to remove. How can I do this without hardcoding it? Note that I can only implement it under C90 because this is an assignment.
Firstly, do not issue wildcard remove commands in a clean: rule (or any other). When the user types make clean, they could lose files that is not even under the control of the build system. Your build system should only clean what it builds. (That is, at least with the "regular strength" clean target; here we should acknowledge the widespread practice of providing additional targets that remove editor backup files, tags, generated configuration materials and whatnot.)
The object files should be listed in some variable like $(OBJS) and should be removed using rm -f ... $(OBJS).
If a certain object file is not to be removed, you can divide things:
DISPOSABLE_OBJS = ...
PRECIOUS_OBJS = ...
OBJS = $(DISPOSABLE_OBJS) $(PRECIOUS_OBJS)
Then in the clean rule, just remove the $(DISPOSABLE_OBJS).
In GNU Make you have additional flexibilities, like rm -f $(filter-out keep1.o keep2.o,$(OBJS)) to remove all $(OBJS) except keep1.o and keep2.o.

Avoid regenerating files that won't change

I have a Makefile with several rules of this form
protolist.c: $(PROTOCOLS) Makefile src/genmodtable.sh
$(SHELL) $(srcdir)/src/genmodtable.sh \
$# $(filter-out %Makefile %genmodtable.sh, $^)
As the name implies, protolist.c winds up containing a list of all the "protocols" defined by the .c files in $(PROTOCOLS). The contents of this file do formally depend on everything in $(PROTOCOLS), the Makefile, and the generator script, but it's very rare for the file to actually change when one of those .c files is edited. Therefore, genmodtable.sh is coded to not change the timestamp of protolist.c if it's not going to make any change to its contents. This causes Make to skip rebuilding protolist.o and its dependencies when it's not really necessary.
That all works fine; the problem is that, because protolist.c now appears to be out of date with respect to its dependencies, Make thinks it has to try to regenerate protolist.c on every run. This isn't a performance issue -- the script is very fast -- but it is confusing behavior. I dimly recall an idiom, involving timestamp files, that could be used to stop Make from doing this, but I have not been able to reconstruct it or find it described anywhere. Does anyone know what it is?
(Also if anyone can suggest how to get rid of that silly $(filter-out ...) construct, that would be helpful, as that is the only GNUmakeism in this Makefile.)
This appears similar to an issue with Fortran programming and make, involving the files generated when compiling a module. (Not relevant, other than that is where I picked up how to do this.)
What you want is have make compare the timestamp of protolist.o to the timestamp of protolist.c, which remains 'old', and make the decision to run the recipe for protolist.c, depending on the timestamp of, well, a timestamp file, which gets updated each time the recipe is run.
In order to make this work, you have to link the two together with an empty rule.
protolist.o: protolist.c
[...]
protolist.c: protolist.c.time ;
protolist.c.time: $(PROTOCOLS) Makefile src/genmodtable.sh
$(SHELL) $(srcdir)/src/genmodtable.sh \
protolist.c $(filter-out %Makefile %genmodtable.sh, $^)
touch protolist.c.time
In my own makefiles, I have to declare the timestamp files as prerequisites of the special target .PRECIOUS, to prevent make from deleting them, but I'm using pattern rules; I'm not 100% sure, but I think this isn't necessary when using explicit rules, like here.
To avoid the $(filter-out ...) construct, can you not simply replace it with $(PROTOCOLS)?
(Although, personally, I would stick to Paul's First Rule of Makefiles: Don't hassle with writing portable makefiles, use a portable make instead.)

Resources