aboutsummaryrefslogtreecommitdiffstats
path: root/guides/source/asset_pipeline.md
diff options
context:
space:
mode:
Diffstat (limited to 'guides/source/asset_pipeline.md')
-rw-r--r--guides/source/asset_pipeline.md9
1 files changed, 6 insertions, 3 deletions
diff --git a/guides/source/asset_pipeline.md b/guides/source/asset_pipeline.md
index 84cda9222e..dd018c0da8 100644
--- a/guides/source/asset_pipeline.md
+++ b/guides/source/asset_pipeline.md
@@ -124,19 +124,22 @@ with a built-in helper. In the source the generated code looked like this:
The query string strategy has several disadvantages:
1. **Not all caches will reliably cache content where the filename only differs by
-query parameters**
+query parameters**
+
[Steve Souders recommends](http://www.stevesouders.com/blog/2008/08/23/revving-filenames-dont-use-querystring/),
"...avoiding a querystring for cacheable resources". He found that in this
case 5-20% of requests will not be cached. Query strings in particular do not
work at all with some CDNs for cache invalidation.
-2. **The file name can change between nodes in multi-server environments.**
+2. **The file name can change between nodes in multi-server environments.**
+
The default query string in Rails 2.x is based on the modification time of
the files. When assets are deployed to a cluster, there is no guarantee that the
timestamps will be the same, resulting in different values being used depending
on which server handles the request.
-3. **Too much cache invalidation**
+3. **Too much cache invalidation**
+
When static assets are deployed with each new release of code, the mtime
(time of last modification) of _all_ these files changes, forcing all remote
clients to fetch them again, even when the content of those assets has not changed.