Scribl is my idea for a system for profiling web servies. My initial hope is to be able to use it with memcached; if that's successful it can be used in more web-stack services.
You are strongly recommended to look in the examples directory, which gives a
concrete example demonstrating how one would use scribl in a real program, and
why it's useful.
Currently scribl is in a pre-alpha state, and is only useful for building trivial demonstration programs.
- Minimal dependencies -- just glib
- Full thread-safety
- Only floating point (double precision) counters are supported
- Time-based threaded serialization
- Performance is not a goal
Scribl depends only on glib, a low-level cross-platform C library. Since scribl relies heavily on glib to implement platform specific routines, it should be pretty portable. In particular, it should compile and run on OS X and Windows (although the developer only has access to Linux for now).
To build scribl you'll also want to have pkg-config installed, although strictly speaking this is not necessary. On Fedora, you'll want the following packages:
- gcc
- glib2-devel
- make
- pkgconfig
On Ubuntu (as of 9.04), the only package you should need to install is
libglib2.0-dev. Everything else should be installed as part of the regular
desktop installation.
To compile and build scribl, just invoke make.
These are currently being tracked in GitHub, see http://github.com/eklitzke/scribl/issues