cache_dir doesn't do what you think it does

[ju5t]

From the README:

This caches the downloaded file in an intermediate directory to avoid repeatedly downloading it. This uses the timestamping (-N) and prefix (-P) wget options to only re-download if the source file has been updated.

   wget::fetch { 'https://tool.com/downloads/tool-1.0.tgz':
     destination => '/tmp/',
     cache_dir   => '/var/cache/wget',
   }

But if you specify cache_dir it actually keeps downloading the file over and over again.

This was introduced in commit back in 2015 from @mirthy (023ed15):

if $redownload == true or $cache_dir != undef  {

Ref: #L62

Is this for any particular reason? Apparently using cache_dir is the only way to set additional parameters on our files such as mode so we prefer to remove the $cache_dir != undef.

I don't mind raising a pull request for it but I wonder what we may break. Can someone shed a light on this perhaps?

[ju5t]

Right, things are finally falling into place now. It actually doesn't download the file again. It only looks like it because it updates the cache and Puppet sees it as a 'change'.

I'm not sure what would be the best way to work around it with Puppet aside from changing wget::fetch into a Ruby function.