Faramarz Jalalian wrote:
> Sriram Natarajan wrote:
> > - You need to collapse the delta *wx redelget* (some files are not
> > done while some looks ok)
> Done and updated the webrev. Please let me know if there are still issues.
> > -  Does your build dump core ?
> Not really, but I thought it's always safe to do what other people are
> doing. Is it already done at a higher level makefile?
> >
> > +
> > +VER=cmake-2.6.1
> > +PREFIX=/usr
> > +
> > +include ../Makefile.cmd
> > +
> > +all: $(VER)/Makefile
> > +    (cd $(VER); env \
> > +        CC="$(CC)" \
> > +        CXX="$(CXX)" \
> > +        "LD_OPTIONS=-M $(SRC)/cmd/mapfile_noexstk -L/usr/lib
> > -R/usr/lib" \
> > +        PATH=$(SFW_PATH) \
> > +        MAKE=$(GMAKE) \
> > +        $(GMAKE) all)
> > +    @find . -type f -name core -exec rm -f {} \;
> > +
> >
> >
> > If this is a cut and paste code, then you don't want this.
> Are you referring to the "@find ..." part or there are other things that
> should be removed such as setting env. vars. BTW,
> I've already removed some of the variables.
> >
> > -  these days, every one is being asked to have 'env -' within their
> > Makefile.
> Should I change the 'env' to 'env -'? I thought that since we're doing
> this prior to invoking nightly, then it should not be needed.

In theory it shouldn't be needed but in real life at least these things
happen:
- "configure"/"autoconf" randomly start/stop depending on environment
variables when people do updates of the code or autoconf which causes
the build to fail in some environments
- People add "export" statements to the interactive shell startup files
(/etc/ksh.kshrc and ~/.kshrc for ksh88/ksh93, /etc/bash.bashrc and
~/.bashrc for bash2/bash3) which may blow-up a build started from within
"bldenv" environment (such "export" statements belong into /etc/profile
and ~/.profile but we can't prevent people from doing it anyway)
- People forget to use "env - ..." to run "nightly".
- ... <insert-more-stuff>

In the past these issues were handled on a case-by-case basis via "unset
OFFENDING_VARIABLE_NAME" each time a breakage occurs - which usually
needs weeks to fix (leaving other people dead in the water until the
issue is fixed) and just eats-up engineering time. That's why I'm now
asking to use "env -" before "configure" and "make"/"gmake" to kill the
root of the problem (which is random environment variables "bleeding
through" into  "configure"/"make"/"gmake") since this _much_ easier to
handle than the endless "add another unset statement" stuff. 
"configure"/"make"/"gmake" should really only see the environment
variables it needs and not random ones.

----

Bye,
Roland

-- 
  __ .  . __
 (o.\ \/ /.o) roland.mainz at nrubsig.org
  \__\/\/__/  MPEG specialist, C&&JAVA&&Sun&&Unix programmer
  /O /==\ O\  TEL +49 641 3992797
 (;O/ \/ \O;)

Reply via email to