Re: makefile and TMPDIR interaction

From: Bill Seivert (seivert_at_pcisys.net)
Date: 02/26/05

  • Next message: Michael B Allen: "Errno Identifer Comparison By System"
    Date: Fri, 25 Feb 2005 20:35:43 -0700
    
    

    Irrational Number wrote:
    > Heny Townsend wrote:
    >
    >
    >>Irrational Number wrote:
    >>
    >>
    >>>In my GNUmakefile, I include a GNUconfigure
    >>>file that contains a call to define TMPDIR = ii_files.
    >>>
    >>>However, this causes a clash, even though the
    >>>GNUconfigure TMPDIR supercedes the other
    >>>definition.
    >>
    >>This is not clearly expressed. If one supersedes the other, there isn't
    >>a clash. Which is it? Do you mean that the setting in .cshrc wins even
    >>though you'd expect the other one to win?
    >
    >
    > Here's where my confusion reigns. The TMPDIR
    > defined in my GNUconfigure file should supersede.
    > When I put a print statement in, that is clearly the
    > case. However, if I define a TMPDIR in my .cshrc,
    > even though the GNUmakefile thinks my TMPDIR
    > is the one in the GNUconfigure, the compilation
    > dies. If I do not define a TMPDIR in my .cshrc,
    > the compilation works. Even if the TMPDIR in my
    > .cshrc gets superseded...
    >
    >
    >>One of the (many) problems with csh is that the .cshrc file is sourced
    >>for every new csh process. If make forks a process and that process is
    >>(or subsequently runs) the csh, the .cshrc will be sourced and the old
    >>TMPDIR will be lost.
    >
    >
    > We use Bourne shell in the Makefiles and every
    > script being called is protected with #!/bin/csh -f,
    > so the .cshrc is not being sourced.
    >
    > We're not "losing" the TMPDIR definition in a
    > bad way. It's being superseded by the definition
    > in the GNUconfigure properly, but the compilation
    > just doesn't work if I have TMPDIR defined in
    > my .cshrc...
    >
    > -- Anita --
    >
    >

    Another option would be to move the TMPDIR definition from
    your .cshrc to your .login. That would avoid the .cshrc
    call from almost anywhere clobbering the make version.

    Bill Seivert


  • Next message: Michael B Allen: "Errno Identifer Comparison By System"

    Relevant Pages

    • Re: makefile and TMPDIR interaction
      ... > defined in my GNUconfigure file should supersede. ... However, if I define a TMPDIR in my .cshrc, ... > even though the GNUmakefile thinks my TMPDIR ... > is the one in the GNUconfigure, the compilation ...
      (comp.unix.programmer)
    • Re: makefile and TMPDIR interaction
      ... > in place of the compiler that print out the value of $TMPDIR. ... > value to something different from either the cshrc value or the configure ... > compilation actually gets called. ... the .cshrc and "ii_files" in the Makefile. ...
      (comp.unix.programmer)
    • Re: makefile and TMPDIR interaction
      ... However, if I define a TMPDIR in my .cshrc, ... Others have warned you that csh is nasty stuff. ... that it's getting re-run from inside, effectively superseding the superseded ...
      (comp.unix.programmer)
    • Re: makefile and TMPDIR interaction
      ... >> In my GNUmakefile, I include a GNUconfigure ... >> However, this causes a clash, even though the ... Do you mean that the setting in .cshrc wins even ... The TMPDIR ...
      (comp.unix.programmer)