[vox-tech] Make with _possibly_ cascading dependencies

[Ken Bloom]

On Thursday 03 May 2007 05:42:37 pm Bill Kendrick wrote:
> This isn't Linux-related, but is make-related ("nmake",
> specificlaly), so I thought I'd ask here. :^)
>
> In the mobile phone software world, you often end up building
> numerous versions of the same app, to address differences between
> handsets (e.g., use different art assets, depending on the screen
> sizes; handset screens differ wildly in size and aspect ratio).
>
> An application I'm working on has had the same data shared across all
> builds (let's name a few of the builds "M", "L" and "XL", for
> example). In my project, I have a data file that gets used during the
> build process.  We now need to be able to use _build-specific_
> versions of this file, for some builds.  (Honestly, it's insanely
> more complicated than all this, but I'm trying to boil it down, for
> consumption :^) )
>
> Example:
>
>   foo.xyz : DATA\foo.in
>   	process DATA\foo.in > foo.xyz
>
> Now, I want to be able to check whether there's a build-specific
> dependency. If so, I want to use it, rather than the 'default' one.
>
> e.g.:
>
>   foo.xyz : DATA\$(SIZE)\foo.in    # Where SIZE is "M", "L" or "XL"
>   	process DATA\$(SIZE)\foo.in > foo.xyz
>
> If and only if it exists... otherwise, use the default one, as shown
> above.
>
>
> My first idea is to just junk the dependency part of the target...
> when I make my builds, I'm always cleaning everything first, anyway. 
> *sigh*
>
> e.g.:
>
>   foo.xyz :
>   	# Do the default first:
>   	process DATA\foo.in > foo.xyz
>   	# Then try to override it with a build-specific one, if it exists:
>   	if EXISTS DATA\$(SIZE)\foo.in   process DATA\$(SIZE)\foo.in >
> foo.xyz
>
>
> An even worse variation on the above is to generate a temp file
> first:
>
>   foo.xyz : DATA\foo.in.temp
>   	process DATA\foo.in.temp > foo.xyz
>
>   foo.in.temp :
>   	# Copy the default file into a temp file, first
>   	copy DATA\foo.in DATA\foo.in.temp
>   	# Overwrite the temp file with a build-specific one, if it exists:
>   	copy DATA\$(SIZE)\foo.in DATA\foo.in.temp
>
> Ideas?

Does nmake support pattern rules like GNU make? If so, and your data is 
suitably organized for a pattern rule, then you can just use those.

[bloom at cat-in-the-hat test]$ cat Makefile
all: test.a
%.a: %.b
	cp $< $@
%.a: %.c
	cp $< $@
[bloom at cat-in-the-hat test]$ echo c > test.c
[bloom at cat-in-the-hat test]$ make
cp test.c test.a
[bloom at cat-in-the-hat test]$ rm test.*
[bloom at cat-in-the-hat test]$ echo b > test.b
[bloom at cat-in-the-hat test]$ make
cp test.b test.a
[bloom at cat-in-the-hat test]$ rm test.*
[bloom at cat-in-the-hat test]$ echo c > test.c
[bloom at cat-in-the-hat test]$ echo b > test.b
[bloom at cat-in-the-hat test]$ make
cp test.b test.a

Just make sure the most general rule is last.


-- 
Ken Bloom. PhD candidate. Linguistic Cognition Laboratory.
Department of Computer Science. Illinois Institute of Technology.
http://www.iit.edu/~kbloom1/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
Url : http://lists.lugod.org/pipermail/vox-tech/attachments/20070503/7603887a/attachment.pgp