Java/parellel build(?) related bootstrap failure

Java/parellel build(?) related bootstrap failure

[Gerald Pfeifer]

...seen on i386-unknown-freebsd4.8:

  /files/pfeifer/OBJ-0329-1255/gcc/gcj
         -B/files/pfeifer/OBJ-0329-1255/i386-unknown-freebsd4.8/libjava/
         -B/files/pfeifer/OBJ-0329-1255/gcc/ --encoding=UTF-8
         -Wno-deprecated -C -g -classpath ''
         -bootclasspath /files/pfeifer/OBJ-0329-1255/i386-unknown-freebsd4.8/libjava:/sw/test/gcc/cvs/libjava \
         -d /files/pfeifer/OBJ-0329-1255/i386-unknown-freebsd4.8/libjava /sw/test/gcc/cvs/libjava/java/util/Vector.java
  /files/pfeifer/OBJ-0329-1255/gcc/gcj
          -B/files/pfeifer/OBJ-0329-1255/i386-unknown-freebsd4.8/libjava/
          -B/files/pfeifer/OBJ-0329-1255/gcc/ --encoding=UTF-8
          -Wno-deprecated -C -g -classpath ''
          -bootclasspath /files/pfeifer/OBJ-0329-1255/i386-unknown-freebsd4.8/libjava:/sw/test/gcc/cvs/libjava \
          -d /files/pfeifer/OBJ-0329-1255/i386-unknown-freebsd4.8/libjava /sw/test/gcc/cvs/libjava/java/util/WeakHashMap.java
  gmake[2]: *** No rule to make target `java/util/prefs/NodeChangeListener.java', needed by `java/util/prefs/NodeChangeListener.class'.  Stop.
  gmake[2]: *** Waiting for unfinished jobs....
  gmake[2]: Leaving directory `/files/pfeifer/OBJ-0329-1255/i386-unknown-freebsd4.8/libjava'
  gmake[1]: *** [all-target-libjava] Error 2

I wonder why nobody else is seeing this.

This is with a 6-way parallel make on a 2-processor box.

Gerald

Re: Java/parellel build(?) related bootstrap failure

[Andreas Jaeger]

Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at> writes:

> ...seen on i386-unknown-freebsd4.8:
>
>   /files/pfeifer/OBJ-0329-1255/gcc/gcj
>          -B/files/pfeifer/OBJ-0329-1255/i386-unknown-freebsd4.8/libjava/
>          -B/files/pfeifer/OBJ-0329-1255/gcc/ --encoding=UTF-8
>          -Wno-deprecated -C -g -classpath ''
>          -bootclasspath /files/pfeifer/OBJ-0329-1255/i386-unknown-freebsd4.8/libjava:/sw/test/gcc/cvs/libjava \
>          -d /files/pfeifer/OBJ-0329-1255/i386-unknown-freebsd4.8/libjava /sw/test/gcc/cvs/libjava/java/util/Vector.java
>   /files/pfeifer/OBJ-0329-1255/gcc/gcj
>           -B/files/pfeifer/OBJ-0329-1255/i386-unknown-freebsd4.8/libjava/
>           -B/files/pfeifer/OBJ-0329-1255/gcc/ --encoding=UTF-8
>           -Wno-deprecated -C -g -classpath ''
>           -bootclasspath /files/pfeifer/OBJ-0329-1255/i386-unknown-freebsd4.8/libjava:/sw/test/gcc/cvs/libjava \
>           -d /files/pfeifer/OBJ-0329-1255/i386-unknown-freebsd4.8/libjava /sw/test/gcc/cvs/libjava/java/util/WeakHashMap.java
>   gmake[2]: *** No rule to make target `java/util/prefs/NodeChangeListener.java', needed by `java/util/prefs/NodeChangeListener.class'.  Stop.
>   gmake[2]: *** Waiting for unfinished jobs....
>   gmake[2]: Leaving directory `/files/pfeifer/OBJ-0329-1255/i386-unknown-freebsd4.8/libjava'
>   gmake[1]: *** [all-target-libjava] Error 2
>
> I wonder why nobody else is seeing this.
>
> This is with a 6-way parallel make on a 2-processor box.

I see this the whole time on the rtlopt-branch and never investigated.
It works for me (2p box, 4way parallel) on all other branches.

Andreas
-- 
 Andreas Jaeger
  SuSE Labs aj@suse.de
   private aj@arthur.inka.de
    http://www.suse.de/~aj

Re: Java/parellel build(?) related bootstrap failure

[Andreas Jaeger]

Andreas Jaeger <aj@suse.de> writes:

> Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at> writes:
>
>> ...seen on i386-unknown-freebsd4.8:
>>
>>   /files/pfeifer/OBJ-0329-1255/gcc/gcj
>>          -B/files/pfeifer/OBJ-0329-1255/i386-unknown-freebsd4.8/libjava/
>>          -B/files/pfeifer/OBJ-0329-1255/gcc/ --encoding=UTF-8
>>          -Wno-deprecated -C -g -classpath ''
>>          -bootclasspath /files/pfeifer/OBJ-0329-1255/i386-unknown-freebsd4.8/libjava:/sw/test/gcc/cvs/libjava \
>>          -d /files/pfeifer/OBJ-0329-1255/i386-unknown-freebsd4.8/libjava /sw/test/gcc/cvs/libjava/java/util/Vector.java
>>   /files/pfeifer/OBJ-0329-1255/gcc/gcj
>>           -B/files/pfeifer/OBJ-0329-1255/i386-unknown-freebsd4.8/libjava/
>>           -B/files/pfeifer/OBJ-0329-1255/gcc/ --encoding=UTF-8
>>           -Wno-deprecated -C -g -classpath ''
>>           -bootclasspath /files/pfeifer/OBJ-0329-1255/i386-unknown-freebsd4.8/libjava:/sw/test/gcc/cvs/libjava \
>>           -d /files/pfeifer/OBJ-0329-1255/i386-unknown-freebsd4.8/libjava /sw/test/gcc/cvs/libjava/java/util/WeakHashMap.java
>>   gmake[2]: *** No rule to make target `java/util/prefs/NodeChangeListener.java', needed by `java/util/prefs/NodeChangeListener.class'.  Stop.
>>   gmake[2]: *** Waiting for unfinished jobs....
>>   gmake[2]: Leaving directory `/files/pfeifer/OBJ-0329-1255/i386-unknown-freebsd4.8/libjava'
>>   gmake[1]: *** [all-target-libjava] Error 2
>>
>> I wonder why nobody else is seeing this.
>>
>> This is with a 6-way parallel make on a 2-processor box.
>
> I see this the whole time on the rtlopt-branch and never investigated.
> It works for me (2p box, 4way parallel) on all other branches.

Investigated now and found that the prefs directory was not on the
branch - I'm merging it now.

Gerald, perhaps you also miss the directory?  Run cvs update -d to get
new directories,

Andreas
-- 
 Andreas Jaeger
  SuSE Labs aj@suse.de
   private aj@arthur.inka.de
    http://www.suse.de/~aj

Re: Java/parellel build(?) related bootstrap failure

[Gerald Pfeifer]

On Sun, 30 Mar 2003, Andreas Jaeger wrote:
> Gerald, perhaps you also miss the directory?  Run cvs update -d to get
> new directories,

Argl, it seems that one tester was indeed lacking -d to avoid getting
gcc/ada. Sorry folks!

Now, isn't there _some_ way to use cvs update -d and still being able to
avoid certain "new" directories like /gcc/ada?

Gerald
-- 
Gerald "Jerry"   pfeifer@dbai.tuwien.ac.at   http://www.pfeifer.com/gerald/

Re: Java/parellel build(?) related bootstrap failure

[Andrew Haley]

Gerald Pfeifer writes:
 > On Sun, 30 Mar 2003, Andreas Jaeger wrote:
 > > Gerald, perhaps you also miss the directory?  Run cvs update -d to get
 > > new directories,
 > 
 > Argl, it seems that one tester was indeed lacking -d to avoid getting
 > gcc/ada. Sorry folks!
 > 
 > Now, isn't there _some_ way to use cvs update -d and still being able to
 > avoid certain "new" directories like /gcc/ada?

You perhaps need a module that checks out an approopriate subset of
gcc.  It should be quite easy to create one.

Andrew.

Re: Java/parellel build(?) related bootstrap failure

[Gerald Pfeifer]

On Mon, 31 Mar 2003, Andrew Haley wrote:
> You perhaps need a module that checks out an approopriate subset of
> gcc.  It should be quite easy to create one.

Thanks for the hint.  Unfortunately, modules don't work nicely with
`cvs update` (one of the design bugs of CVS, that it after checkout,
it simpley ignores modules).

In case someone else has a similar problem, here is what I wound up
using now:

  cvs -q $CVS_EXT update -Pd -I\! -I CVS libjava \
  &&  cvs -q $CVS_EXT update -P -I\! -I CVS
  if [ ! $? ]; then
      echo "Error updating. Build aborted."
      exit 1
  fi
  contrib/gcc_update --touch

Gerald
-- 
Gerald "Jerry"   pfeifer@dbai.tuwien.ac.at   http://www.pfeifer.com/gerald/

Re: Java/parellel build(?) related bootstrap failure

[Gerald Pfeifer]

On Mon, 31 Mar 2003, Andrew Haley wrote:
> You perhaps need a module that checks out an approopriate subset of
> gcc.  It should be quite easy to create one.

Thanks for the hint.  Unfortunately, modules don't work nicely with
`cvs update` (one of the design bugs of CVS, that it after checkout,
it simpley ignores modules).

In case someone else has a similar problem, here is what I wound up
using now:

  cvs -q $CVS_EXT update -Pd -I\! -I CVS libjava \
  &&  cvs -q $CVS_EXT update -P -I\! -I CVS
  if [ ! $? ]; then
      echo "Error updating. Build aborted."
      exit 1
  fi
  contrib/gcc_update --touch

Gerald
-- 
Gerald "Jerry"   pfeifer@dbai.tuwien.ac.at   http://www.pfeifer.com/gerald/

Re: Java/parellel build(?) related bootstrap failure

[Daniel Jacobowitz]

On Wed, Apr 02, 2003 at 12:54:00AM +0200, Gerald Pfeifer wrote:
> On Mon, 31 Mar 2003, Andrew Haley wrote:
> > You perhaps need a module that checks out an approopriate subset of
> > gcc.  It should be quite easy to create one.
> 
> Thanks for the hint.  Unfortunately, modules don't work nicely with
> `cvs update` (one of the design bugs of CVS, that it after checkout,
> it simpley ignores modules).
> 
> In case someone else has a similar problem, here is what I wound up
> using now:
> 
>   cvs -q $CVS_EXT update -Pd -I\! -I CVS libjava \
>   &&  cvs -q $CVS_EXT update -P -I\! -I CVS
>   if [ ! $? ]; then
>       echo "Error updating. Build aborted."
>       exit 1
>   fi
>   contrib/gcc_update --touch

I generally just use checkout again:
  ln -s . gcc
  cvs co gcc-core gcc-java (or whatever)
  rm -f gcc

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

Re: Java/parellel build(?) related bootstrap failure

[Andrew Haley]

Gerald Pfeifer writes:
 > On Mon, 31 Mar 2003, Andrew Haley wrote:
 > > You perhaps need a module that checks out an approopriate subset of
 > > gcc.  It should be quite easy to create one.
 > 
 > Thanks for the hint.  Unfortunately, modules don't work nicely with
 > `cvs update` (one of the design bugs of CVS, that it after checkout,
 > it simpley ignores modules).

You just checkout the module again, on top of the old one.  It works
fine, just like cvs update -d, but without the directories you don't
need.

Andrew.

Re: Java/parellel build(?) related bootstrap failure

[Tom Tromey]

>>>>> "Gerald" == Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at> writes:

Gerald>   gmake[2]: *** No rule to make target `java/util/prefs/NodeChangeListener.java', needed by `java/util/prefs/NodeChangeListener.class'.  Stop.

Gerald> I wonder why nobody else is seeing this.

Does that file exist in your source tree?
If not, did you `cvs update -d'?
If you're working on a branch, did the file get merged to the branch?
If none of those, are you using GNU make?

If it isn't any of those, then I don't know what it is offhand.
We'll have to debug.

Tom