Issue 3019: Python3a5 compile failing due to high memory usage (original) (raw)

I have tried building Python3a5 today using the tarball offered on your internet site. The process './python -E setup.py build' keeps on allocating my memory using more than 80% of my RAM, I am certain that it would have used more if I had not cancelled it. When I use ltrace to track the system calls and put it into the Makefile, the resource usage still raises but much slower, if I use ltrace and supply it a PID, the memory usage raises as fast as normal. I think this is due to something in ltrace, it's understandable because it produced a file of more than 600MB.

I attached an excerpt of the output of ltrace, showing you how the system calls seem to repeat, and the output of './configure', 'make', 'uname -a', 'lspci' and 'lsmod' so you have some data about my system. I also got a screenshot showing the output of htop, make and strace together: http://img516.imageshack.us/img516/2275/screenshot1ab8.png. Maybe the output of two different programs(ltrace and strace) will help you too. I tried out building the current SVN py3k branch, resulting in a similar problem. PS: It would help if you allowed uploading of multiple files to bug reports