pcp
[Top] [All Lists]

fche/dev patch RFC: check-statis

To: pcp developers <pcp@xxxxxxxxxxx>
Subject: fche/dev patch RFC: check-statis
From: "Frank Ch. Eigler" <fche@xxxxxxxxxx>
Date: Wed, 19 Mar 2014 23:17:20 -0400
Delivered-to: pcp@xxxxxxxxxxx
User-agent: Mutt/1.4.2.2i
Hi -

Please ponder and backfeed re. this fresh commit on fche/dev,
initially motivated by nathans' rhel5 build problems tonight.  (The
commit may be more easily grokked with "git diff --word-diff").  I
don't have access to MacOS etc. to test it on, but I don't think that
logic is affected.

commit db04ebe4297145a2b365ecd23b39130b2baef636

    libpcp/src/check-statics: simplify, pessify
    
    check-statics has been in some cases overly and in others underly
    sensitive.
    
    In the overly sensitive category, it's too sensitive to compilers'
    choice of section placement and for static data and suffixing due to
    function-static / optimization.  It seems safe to accept the symbols
    with any of the usual prefixes and suffixes.
    
    In the underly sensitive category, check-statics could exit with a
    success rc in a variety of conditions, such as being interrupted, or a
    failure of an inferior command, or even a script syntax error.  Let's
    flip the $sts default around and add a 'set -e' to become very
    conservative with respect to environmental catastrophes.

<Prev in Thread] Current Thread [Next in Thread>
  • fche/dev patch RFC: check-statis, Frank Ch. Eigler <=