Not out in the field yet but I'm thinking that user agent strings may not be the only "helpful" header.
Performance wise, this shouldn't be a penalty as I believe the request headers are cached after reading once so calling HttpHeaders for the User-Agent repeatedly in the methods shouldn't hurt. On Thu, Sep 24, 2009 at 3:34 AM, Baram, Eliezer <[email protected]> wrote: > Sounds good, it makes it more generic. I'm curious did you run into a use > case that need this? > > -----Original Message----- > From: Bryant Luk [mailto:[email protected]] > Sent: Thursday, September 24, 2009 5:33 AM > To: [email protected] > Subject: [DISCUSS] MediaTypeMapper changes > > Hi, > > I would like to propose changes to MediaTypeMapper. > > First, I think we should make it externally configurable as right now, > it's internal. > > Second, I was hoping to change the interface slightly for determining > if there's a match. Basically, allow the media type mapper to work > with any incoming request headers instead of just user-agents. > > In MediaTypeMapper: > > public MediaType mapOutputMediaType(MediaType responseMediaType, > String userAgent) > > would change to: > > public MediaType mapOutputMediaType(MediaType responseMediaType, > HttpHeaders requestHeaders) > > and: > > In MappingRecord: > boolean match(String userAgent, MediaType responseMediaType); > would change to: > boolean match(HttpHeaders requestHeaders, MediaType responseMediaType); > > I imagine a future use case may be that the user agent header may not > be the only header that has to be taken into consideration and maybe a > user agent match can't be done with just a starting with. > > Thoughts? > > -- > > - Bryant Luk > -- - Bryant Luk
