At the Linked Data preconference at Code4Lib 2009 a couple of weeks ago, I learned about the rdflib Python library. Naturally, I wanted to install the library so I could mess with it on my own. That proved to be a little problematic, though. Following the installation instructions, I typed:
$ easy_install -U "rdflib>=2.4,<=3.0a"
And it very helpfully told me (among other things):
gcc -pthread -fno-strict-aliasing -DNDEBUG -g -O2 -Wall -Wstrict-prototypes -fPIC -I/usr/include/python2.5 -c src/bison/SPARQLParser.c -o build/temp.linux-i686-2.5/src/bison/SPARQLParser.o src/bison/SPARQLParser.c:7:20: error: Python.h: No such file or directory
As it turns out, I should have had python-dev and build-essential already installed. Being new to Linux, I did not know. Thanks, BenO, for helping me out with this.
The complete installation instructions:
$ apt-get install python-dev $ apt-get install build-essential $ easy_install "rdflib==2.4.0"
This is one thing that more and more distros have been doing over the past couple years that bugs me a bit. I don’t know why build-essential isn’t part of the default install. The lovely idea that the package manager will be able to always use a binary or never need to use c headers seems pretty rare, except maybe for the most mom & pop users.
And not including the -devel packages by default? I guess they’re trying to save bandwidth.
I suppose there could be a security concern, but really, how many worms/virus/etc re-compile themselves?
Still, everytime I find myself working with a new linux system, I run into exactly this problem. Annoying, but I guess it isn’t too bad. Thank goodness for the Internet and Google.