Static Asset Compilation

Fairly standard stuff. If you're going to have a title with "you're doing it wrong" you should have some unique insight to support your dramatised title.

With such a complicated system I think you are missing out on the most signifcant speed optimization technique; reducing http requests.<p>For reference: <a href="http://developer.yahoo.com/blogs/ydn/posts/2007/04/rule_1_make_few/" rel="nofollow">http://developer.yahoo.com/blogs/ydn/posts/2007/04/rule_1_ma...</a><p>It's laudable that you are paying attention to caching, but you don't compile all your files in to one file. It seems like you could pick up a lot ground here by at least concating all css into one file, and js in to one file.<p>Also you could load jquery from the Google AJAX API endpoint. That way the users has a higher chance of having already loaded jQuery.<p>Also using the same CSS/JS products across multiple pages would help.

You don't need to rename the files. As of a few months ago, you can configure CloudFront to take query strings into account when caching, so you can simply link to the file as normal but append "?&#60;your_hash_here&#62;" to the filename. (I actually prefer using the last modified timestamp over a hash.) IMHO, this is better because it requires less magic on the origin server. And even <i>ancient</i> references to a file (a logo someone hotlinked, for example) will still render rather than 404, so long as the name hasn't changed. No need to keep tons of old revisions of files around.

Could someone say if using HttpGzipStaticModule really helps? Gzipping small static resources on-the-fly should not take down your cpu by much.<p>Surely a nice thing to have, but does it help?

I found that renaming CSS file names using the hash of the contents does not always work because any changes to dependencies (e.g. images) won't always bubble up to the CSS. I forget all of the reasons why it didn't always work, but it I think it had to do with CDN invalidation for files that I could not-rename (e.g index.html).<p>The process I use computes the hash of every file and creates a dependency map then I use the hash of the contents of a file and its dependencies to rename the file.

Can someone with knowledge of both this and rails 3.1 explain the difference. Seems very similar.

Good tips. I've found that <a href="http://pngquant.org" rel="nofollow">http://pngquant.org</a> generates smaller pngs than optipng, but the former is lossy (reduced color palette). I can't tell the difference though.

Aside from minifying javascript, you should probably also consider using Google's Closure Compiler in 'Advanced-Compilation' mode. I believe it does a much better job than traditional minification.<p><a href="https://developers.google.com/closure/compiler/docs/api-tutorial3" rel="nofollow">https://developers.google.com/closure/compiler/docs/api-tuto...</a>

Is putting hash digests in filenames really easier than sending <i>Last-Modified</i> headers in the response, parsing If-<i>Modified-Since</i> headers and returning 304 when applicable, and/or using ETags?<p>I would have thought that most web frameworks do all these things for you automatically by now.

We use the git commit hash of the checkin that is pushed to production as part of the folder structure for our assets. Has worked really well for us.<p>It does mean we use more space on s3 though, but it guarantees we won't miss re-seeding some of the files.

Interesting series <a href="http://autoref.com/blog/2012/09/07/the-tech-behind-autoref-part-1-our-stack/" rel="nofollow">http://autoref.com/blog/2012/09/07/the-tech-behind-autoref-p...</a><p>Full disclosure (bizdev at AutoRef)

Why is it safe to include a subset of the SHA1 digest instead of the whole digest? What's the reasoning behind this? Would it make sense to use a shorter hash (e.g. CRC32) instead if your filenames have to be that short?

You could probably shave a few bytes off your URL's, while achieving the same collision resistance (or alternatively increase the collision resistance in the same number of bytes) if you base64 encoded the hash.

If you're using .net, RequestReduce [1] is an excellent tool for managing your static assets.<p>[1] - <a href="https://github.com/mwrock/RequestReduce" rel="nofollow">https://github.com/mwrock/RequestReduce</a>

Is ImageOptim still a good choice to go with? I really like the simplicity of the GUI.