This fixes it for us. We are using a proc for the asset host. I'm really curious though - what is the use case where you would want to change the fingerprint on an asset just because the asset host changes? If the content of the asset hasn't changed what's the point?

If asset host changes the content also change since the assets url inside
the stylesheet use the full url that contains the configured assets host
On Apr 21, 2014 11:57 AM, "Scott Jacobsen" notifications@github.com wrote:

This fixes it for us. We are using a proc for the asset host. I'm really
curious though - what is the use case where you would want to change the
fingerprint on an asset just because the asset host changes? If the content
of the asset hasn't changed what's the point?

—
Reply to this email directly or view it on GitHub/~https://github.com//pull/140#issuecomment-40952075
.

Okay. I would have expected the fingerprint to change because the content
of the stylesheet has changed. The portion of the fingerprint that is based
on the file content must be calculated before the expanded paths are
inserted into the stylesheet?

On Mon, Apr 21, 2014 at 11:00 AM, Rafael Mendonça França <
notifications@github.com> wrote:

If asset host changes the content also change since the assets url inside
the stylesheet use the full url that contains the configured assets host
On Apr 21, 2014 11:57 AM, "Scott Jacobsen" notifications@github.com
wrote:

This fixes it for us. We are using a proc for the asset host. I'm really
curious though - what is the use case where you would want to change the
fingerprint on an asset just because the asset host changes? If the
content
of the asset hasn't changed what's the point?

Reply to this email directly or view it on GitHub<
/~https://github.com/rails/sprockets-rails/pull/140#issuecomment-40952075>
.

Reply to this email directly or view it on GitHub/~https://github.com//pull/140#issuecomment-40952389
.

Well yes, that's the whole point: this is about the fingerprint of the source file(s), so sprockets knows whether it needs to recompile.

rebased, inline conditional

Just spent quite some time figuring out why assets on two nodes got different fingerprints after a deployment. Thank you for fixing this! Hopefully the fix will be published to Rubygems soon.

Same here. Took me about half a day downtime to get to this. I hope we can expect a release soon :)

Same here. I'm glad this is fixed but I don't care much for the approach. IMHO the digest of identical files should be identical and not depend on where they are hosted. The downside of the current approach is that the files on a staging server will always be different from a production server which would typically use a different asset host.

I understand that this is done as a hack to make sprockets realise the contents of a CSS file can change depending on the asset_host, but can't we think of something more elegant? Detecting whether the asset host is referenced anywhere in the file perhaps?

@rafaelfranca Any chance this could be released in a 2.1.4 version? It's giving us inconsistent assets across multiple machines which is a pretty serious bug IMO.

@braddunbar Released