Add the Accept-Encoding header to the cache key

[trasch]

This is necessary so that clients who request an uncompressed body won't get a gzip'ed result (and vice versa, but that would be less critical).

Note: There might be more headers to check for, because they might change the result body.

[codecov-io]

Current coverage is 100.00%

Merging #1 into master will not affect coverage as of 683e06b

@@            master      #1   diff @@
======================================
  Files            1       1       
  Stmts           21      21       
  Branches         0       0       
  Methods          0       0       
======================================
  Hit             21      21       
  Partial          0       0       
  Missed           0       0       

Review entire Coverage Diff as of 683e06b

Powered by Codecov. Updated on successful CI builds.

[brianreavis]

I worry that clients' Accept-Encoding can be pretty varied, which will bloat the cache. What do you think of making they cache key configurable like below?

lru({key: function(req) {
    return req.z+','+req.x+','+req.y+','+req.layer+','+req.filename+','+req.headers['accept-encoding'];
}});

This way we keep the default behavior for most cases where it doesn't make a difference – yet there's a way to deal with cases like you mention. It's a little clunky but is most configurable. With this sort of setup you could even cut down on cache space like so:

lru({key: function(req) {
    var compressed = /gzip|deflate/.test(req.headers['accept-encoding']) ? 1 : 0;
    return req.z+','+req.x+','+req.y+','+req.layer+','+req.filename+','+compressed;
}});

[trasch]

Good idea, I changed the patch accordingly!

[brianreavis]

Awesome! This is now out: 0.2.0. Really appreciate all the misc fixes / improvements you've been putting in.

[trasch]

You're welcome, and I appreciate all the effort you've already put into Tilestrata! :-)