[uae] Re: CAPS plugin (was 'Compile problems with UAE 0.8.25')
- From: Derek Piper <dcpiper@xxxxxxxxxxx>
- To: uae@xxxxxxxxxxxxx
- Date: Mon, 30 Aug 2004 08:31:14 -0500
Richard Drummond wrote:
when I do 'make', it runs briefly and this error is seen amongst the
output:
make -C tools build68kc
make[2]: Entering directory `/usr/src/uae-0.8.25-20040302/src/tools'
make[2]: *** No rule to make target `build68kc'. Stop.
make[2]: Leaving directory `/usr/src/uae-0.8.25-20040302/src/tools'
make[1]: *** [tools/build68k] Error 2
make[1]: *** Waiting for unfinished jobs....
That's very strange. Look in the src/tools directory. Has the config script
created a symlink from build68k.c to the same file in the src directory?
(Yeah, I know this is a convoluted way of doing things, but I've done it like
this so that the build system supports cross-compiling without having to
re-arrange the directory structure.)
It appears to work, convoluted or no; the symlink is created.
I did notice this:
in src/Makefile:
tools/build68k:
$(MAKE) -C tools build68kc
in src/tools/Makefile:
build68k: build68k.o writelog.o
$(CC) $(LDFLAGS) -o build68k build68k.o writelog.o
So that's obviously why it's failing at that point. The parent Makefile
is calling the wrong label, but since it also makes 'all-recursive' it
hits the src/tools/Makefile which within ITS own 'all' target has
'build68k' correctly, so the binary is built. Next time around the
'tools/build68k' target has been met, so there is no persistent problem.
Let's continue on with the play-by-play, second time around we get to
this stage:
compemu_support.c:15: comptbl.h: No such file or directory
I can see that the binary 'gencomp' was compiled just after it though,
in the tools directory. After this invocation of 'make' has finished,
comptbl.h still does not exist.
Next run through './tools/gencomp' is run, comptbl.h now exists and the
build appears to finish complete.
I probably should have looked at it the other day, I was just in 'huh?'
mode at the time I think. Hopefully that gives some more useful
information as to what is happening.
Hang on a minute . . . that says 'build68kc' not ' build68k.c'. Very strange
indeed. Can you send me a copy of your src/tools/Makefile?
I will happily send you any of the Makefiles, is it okay to spam the
list with them though?
I've not tried building on Woody for a while (and I don't have a Woody
installation at the moment), but I didn't encounter any problems there before
as I recall.
Every time I run make it gets a little bit closer to finishing. Is the
order of the targets incorrect? Why does it need to have to be
'prodded' along like that? 'make' needs to be run 3 times in order for
the build to complete. Is that normal?
Very bizarre.
With the revised configure options, it still exhibits that behaviour. I
have to type 'make' a few times to get it to compile fully. Anyone know
why it does that?
No. :-(
It's not what I would expect at all. If it were broken, it should stay broken
- dammit. Running make again should make no difference.
Without the SDL options above I get a build that at least loads the
kickstart.
Okay. I did find that my problems only came when I had
'--with-sdl-sound' in my options, so I guess I'll leave that out. I
still get sound.
SDL audio generally causes lots of problems if you have an aRts-enabled build
of SDL.
If you leave out the --with-sdl-sound option on Linux, you get OSS sound.
Okay, that's cool. I do only have the libsdl1.2-oss library installed
but it doesn't seem like much of a difference in the result to not use
SDL for sound.
Other related posts:
