Use gzip python library instead of sys call for compressing buildlogs

it's a simple example I wrote to use gzip and it works fine.

import gzip
import tempfile
import os

CONTENT = 'Hello World !!!'

def gzipContent(content):
    """Return the filename of gzipped content."""
    out_file_fd, out_file_name = tempfile.mkstemp(suffix=".gz")
    gz_file = gzip.GzipFile(out_file_name, mode='wb')
    gz_file.write(content)
    gz_file.close()

    return out_file_name

def fileSize(filename):
    """Return the real file size."""
    _file = open(filename, 'r')
    _file.seek(0,2)
    size = _file.tell()
    _file.close()
    return size

def gunzipFile(filename):
    """Uncompress filename and return its content """
    gz_file = gzip.open(filename)
    raw = gz_file.read()
    gz_file.close()
    return raw

if __name__ == '__main__':

    filename = gzipContent(CONTENT)
    print 'Compressed: %s' % (filename)

    gz_size = fileSize(filename)
    print 'Gzip file size: %d' % gz_size

    content = gunzipFile(filename)
    print 'Raw content: %s' % content

    os.remove(filename)

Just so I know, is there any advantage to using the gzip library instead of os.system("gzip")?

Well, it would just make the code more elegant, no other explicit advantage so far.

It is fine by me to keep it as 'wishlist' only, if you want to.

Agreed.

Barry, when reviewing some related code has suggested a short-term solution that might be worth to try:
{{{
    I read the bug and I agree it would be more elegant to use the gzip library,
    but barring that, what happens if the gzip command fails? Would it make more
    sense to use subprocess here to handle potential gzip failures?
}}}

The barry's suggestion is easy to implement and will leave us in a more comfortable situation.

Optional for 1.1.7, may be left for 1.1.8

definitely 1.1.8

RF 4653