[RFA/testsuite/ada] re-implement make clean & distclean

[RFA/testsuite/ada] re-implement make clean & distclean

[Joel Brobecker]

[-- Attachment #1: Type: text/plain, Size: 761 bytes --]

Hello,

The current implementation for make clean requires us that we update
the EXECUTABLES list each time we add a new testcase. It was a vague
copy/paste of some code I probably found in one of the nearby Makefiles...
It was causing me a bit of grief that certain files would not be deleted
when I did a "make clean", in particular the executables produced by
gnatmake. I decided to replace this with an approach we have been using
with great success in our own testsuite for years...

2006-12-31  Joel Brobecker  <brobecker@adacore.com>

        * gdb.ada/Makefile.in (EXECUTABLES): Delete.
        (MISCELLANEOUS): Delete.
        (clean): Re-implement.
        (mostlyclean): Likewise.

Tested on my x86-linux laptop.  Any objection?

Thank you,
-- 
Joel

[-- Attachment #2: gdb.ada.clean.dif --]
[-- Type: text/plain, Size: 1306 bytes --]

Index: Makefile.in
===================================================================
RCS file: /cvs/src/src/gdb/testsuite/gdb.ada/Makefile.in,v
retrieving revision 1.2
diff -u -p -r1.2 Makefile.in
--- Makefile.in	3 Feb 2005 03:58:52 -0000	1.2
+++ Makefile.in	31 Dec 2006 10:00:46 -0000
@@ -1,24 +1,16 @@
 VPATH = @srcdir@
 srcdir = @srcdir@
 
-EXECUTABLES = null_record/null_record fixed_points/fixed_points
-
-MISCELLANEOUS =
-
 all info install-info dvi install uninstall installcheck check:
 	@echo "Nothing to be done for $@..."
 
 clean mostlyclean:
-	-find . -name '*.o' -print | xargs rm -f
-	-find . -name '*.ali' -print | xargs rm -f
-	-find . -name 'b~*.ad[sb]' -print | xargs rm -f
-	-rm -f *~ a.out xgdb *.x *.ci *.tmp
-	-rm -f *~ *.o a.out xgdb *.x *.ci *.tmp
-	-rm -f core core.coremaker coremaker.core corefile $(EXECUTABLES)
-	-rm -f $(MISCELLANEOUS) twice-tmp.c
+	-find . ! \( -name CVS -prune \) ! -type d \
+	  ! -name '*.ad[sb]' ! -name '*.[hc]' ! -name '*.gpr' \
+	  ! -name '*.exp' \
+	  ! -name 'Makefile*' \
+	  -exec rm -f {} \;
+	-find . -name 'b~*.ad[sb]' -exec rm -f {} \;
 
 distclean maintainer-clean realclean: clean
-	-rm -f *~ core
-	-rm -f Makefile config.status config.log
-	-rm -f *-init.exp
-	-rm -fr *.log summary detail *.plog *.sum *.psum site.*
+	-rm -f Makefile

Re: [RFA/testsuite/ada] re-implement make clean & distclean

[Mark Kettenis]

> Date: Sun, 31 Dec 2006 14:07:55 +0400
> From: Joel Brobecker <brobecker@adacore.com>
> 
> Hello,
> 
> The current implementation for make clean requires us that we update
> the EXECUTABLES list each time we add a new testcase. It was a vague
> copy/paste of some code I probably found in one of the nearby Makefiles...
> It was causing me a bit of grief that certain files would not be deleted
> when I did a "make clean", in particular the executables produced by
> gnatmake. I decided to replace this with an approach we have been using
> with great success in our own testsuite for years...
> 
> 2006-12-31  Joel Brobecker  <brobecker@adacore.com>
> 
>         * gdb.ada/Makefile.in (EXECUTABLES): Delete.
>         (MISCELLANEOUS): Delete.
>         (clean): Re-implement.
>         (mostlyclean): Likewise.
> 
> Tested on my x86-linux laptop.  Any objection?

Hmm, seems a bit scary to me.  If I read this correctly, a file named
NOTES would be wiped isn't it?  I'm using a seperate object dir, so I
probably won't care, but others might...

Re: [RFA/testsuite/ada] re-implement make clean & distclean

[Joel Brobecker]

> > 2006-12-31  Joel Brobecker  <brobecker@adacore.com>
> > 
> >         * gdb.ada/Makefile.in (EXECUTABLES): Delete.
> >         (MISCELLANEOUS): Delete.
> >         (clean): Re-implement.
> >         (mostlyclean): Likewise.
> > 
> > Tested on my x86-linux laptop.  Any objection?
> 
> Hmm, seems a bit scary to me.  If I read this correctly, a file named
> NOTES would be wiped isn't it?  I'm using a seperate object dir, so I
> probably won't care, but others might...

The thing is that we very rarely add files other than the exp files
and the source code. If we add a file called NOTES in gdb.ada, then
we can adjust the exclusion list in the find command. Doing it the
other way around is a lot more work.

-- 
Joel

Re: [RFA/testsuite/ada] re-implement make clean & distclean

[Jim Blandy]

Joel Brobecker <brobecker@adacore.com> writes:
>> > 2006-12-31  Joel Brobecker  <brobecker@adacore.com>
>> > 
>> >         * gdb.ada/Makefile.in (EXECUTABLES): Delete.
>> >         (MISCELLANEOUS): Delete.
>> >         (clean): Re-implement.
>> >         (mostlyclean): Likewise.
>> > 
>> > Tested on my x86-linux laptop.  Any objection?
>> 
>> Hmm, seems a bit scary to me.  If I read this correctly, a file named
>> NOTES would be wiped isn't it?  I'm using a seperate object dir, so I
>> probably won't care, but others might...
>
> The thing is that we very rarely add files other than the exp files
> and the source code. If we add a file called NOTES in gdb.ada, then
> we can adjust the exclusion list in the find command. Doing it the
> other way around is a lot more work.

Mistakes in Joel's suggested new procedure would be caught by CVS ---
another point in its favor.

Re: [RFA/testsuite/ada] re-implement make clean & distclean

[Daniel Jacobowitz]

On Sun, Dec 31, 2006 at 06:32:37PM +0400, Joel Brobecker wrote:
> > > 2006-12-31  Joel Brobecker  <brobecker@adacore.com>
> > > 
> > >         * gdb.ada/Makefile.in (EXECUTABLES): Delete.
> > >         (MISCELLANEOUS): Delete.
> > >         (clean): Re-implement.
> > >         (mostlyclean): Likewise.
> > > 
> > > Tested on my x86-linux laptop.  Any objection?
> > 
> > Hmm, seems a bit scary to me.  If I read this correctly, a file named
> > NOTES would be wiped isn't it?  I'm using a seperate object dir, so I
> > probably won't care, but others might...
> 
> The thing is that we very rarely add files other than the exp files
> and the source code. If we add a file called NOTES in gdb.ada, then
> we can adjust the exclusion list in the find command. Doing it the
> other way around is a lot more work.

Except that, like Mark implies, I tend to keep my notes in my working
directory.  If I'm debugging a bunch of testcases and I needed to keep
notes about what was going on, I'd be quite surprised if make clean
deleted my notes!

I don't object too strongly, though, if others like this approach.
I wouldn't put them in that directory anyway, just in gdb/testsuite/.

-- 
Daniel Jacobowitz
CodeSourcery

Re: [RFA/testsuite/ada] re-implement make clean & distclean

[Joel Brobecker]

> Except that, like Mark implies, I tend to keep my notes in my working
> directory.  If I'm debugging a bunch of testcases and I needed to keep
> notes about what was going on, I'd be quite surprised if make clean
> deleted my notes!

Hum, I hadn't thought about that. I keep my own notes in one central
directory - mostly to avoid losing them more than anything, it's hard
to keep track of what is in my hard drive sometimes...

I don't know how else to solve our issue. Maintaining a list of files
to delete is a royal pain :-(.

> I don't object too strongly, though, if others like this approach.
> I wouldn't put them in that directory anyway, just in gdb/testsuite/.

Fortunately, I only suggest this for gdb.ada, so very few people would
be impacted. Perhaps the lack of consistency is a bad thing?

-- 
Joel

Re: [RFA/testsuite/ada] re-implement make clean & distclean

[Daniel Jacobowitz]

On Sun, Dec 31, 2006 at 07:27:43PM +0400, Joel Brobecker wrote:
> > I don't object too strongly, though, if others like this approach.
> > I wouldn't put them in that directory anyway, just in gdb/testsuite/.
> 
> Fortunately, I only suggest this for gdb.ada, so very few people would
> be impacted. Perhaps the lack of consistency is a bad thing?

We have the same problem all over the testsuite, really - no one keeps
the makefiles up to date any more.

-- 
Daniel Jacobowitz
CodeSourcery

Re: [RFA/testsuite/ada] re-implement make clean & distclean

[Joel Brobecker]

> > Fortunately, I only suggest this for gdb.ada, so very few people would
> > be impacted. Perhaps the lack of consistency is a bad thing?
> 
> We have the same problem all over the testsuite, really - no one keeps
> the makefiles up to date any more.

Just to be clear, given the legitimate objections I've heard, I am
withdrawing this patch. I'll see how to clean the Makefile up. Thanks
to everyone for their feedback.

-- 
Joel