Comments on: Real-world Performance Tuning with Callgrind http://blog.josefsson.org/2008/02/27/real-world-performance-tuning-with-callgrind/ Wed, 20 Aug 2008 14:00:18 +0000 http://wordpress.org/?v=2.6 By: Aaron M. Ucko http://blog.josefsson.org/2008/02/27/real-world-performance-tuning-with-callgrind/#comment-4864 Aaron M. Ucko Mon, 03 Mar 2008 19:57:00 +0000 http://blog.josefsson.org/2008/02/27/real-world-performance-tuning-with-callgrind/#comment-4864 Debian's policy is to install shared libraries stripped of debugging symbols, which instead go (at least for reasonably popular libraries) in separate -dbg packages. In this case, installing libgnutls26-dbg or libgnutls13-dbg as appropriate would have gotten you the symbols. Debian’s policy is to install shared libraries stripped of debugging symbols, which instead go (at least for reasonably popular libraries) in separate -dbg packages. In this case, installing libgnutls26-dbg or libgnutls13-dbg as appropriate would have gotten you the symbols.

]]> By: simon http://blog.josefsson.org/2008/02/27/real-world-performance-tuning-with-callgrind/#comment-4858 simon Mon, 03 Mar 2008 14:59:35 +0000 http://blog.josefsson.org/2008/02/27/real-world-performance-tuning-with-callgrind/#comment-4858 Marius, whether that trick works or not probably depends on how the shared/static libraries were compiled. I didn't get debugging symbols with the shared library on my Debian box, so my first try to work around that was to link against the static library. Another alternative would be to build a local copy of a shared GnuTLS library with CFLAGS=-g but that would have been more complicated to explain in a write-up like this. Marius, whether that trick works or not probably depends on how the shared/static libraries were compiled. I didn’t get debugging symbols with the shared library on my Debian box, so my first try to work around that was to link against the static library. Another alternative would be to build a local copy of a shared GnuTLS library with CFLAGS=-g but that would have been more complicated to explain in a write-up like this.

]]> By: Marius Gedminas http://blog.josefsson.org/2008/02/27/real-world-performance-tuning-with-callgrind/#comment-4855 Marius Gedminas Mon, 03 Mar 2008 13:51:31 +0000 http://blog.josefsson.org/2008/02/27/real-world-performance-tuning-with-callgrind/#comment-4855 Nice story. A good illustration of the principles of optimization: first profile, then look for a better algorithm (and only then consider microoptimizations). I didn't know the trick with linking libraries statically to get debug symbols. Nice story. A good illustration of the principles of optimization: first profile, then look for a better algorithm (and only then consider microoptimizations).

I didn’t know the trick with linking libraries statically to get debug symbols.

]]> By: Diego http://blog.josefsson.org/2008/02/27/real-world-performance-tuning-with-callgrind/#comment-4834 Diego Mon, 03 Mar 2008 02:24:47 +0000 http://blog.josefsson.org/2008/02/27/real-world-performance-tuning-with-callgrind/#comment-4834 It's nice to see how you were able to use those tools to find the problem so easily. Thanks for writing about it. It’s nice to see how you were able to use those tools to find the problem so easily. Thanks for writing about it.

]]>