Andres March - [31/Mar/05 02:32 PM ] Subject: Re: OSCache wrong; SiteMesh right? According to Jayson Falkner and his book(s) "Servlets and JavaServer pages; the J2EE Web Tier" we have to change the line to cachedWriter = new PrintWriter(new OutputStreamWriter(getOutputStream(), result.getContentEncoding())); in CacheHttpServletResponseWrapper and we have to add a getter method in ResponseContent public String getContentEncoding() {         return contentEncoding;     } Patrick Lightbody wrote: >Joe's suggestion sounds spot-on. I'm going to modify 2.0.2 and try it >out now. >Subject: Re: OSCache wrong; SiteMesh right? > >I see. There is behavior added in the getOutputStream() and that is >what is being wrapped. The super writer will write to the original >output stream not the caching one. The encoding might be the issue and >is a problem either way. I'll get it fixed this weekend. > >Joe Walnes wrote: > > > >>First thing to do is prove which filter is correct. Try the web-app >>with both filters, then each of them individually, then neither, to >>narrow it down. >> >>I suspect the problem is that in >>CacheHttpServletResponseWrapper.getWriter(), a new PrintWriter is >>created wrapped around the ServletOutputStream, but it's not taking >>into account the encoding. >> >>I reckon the problem might go away if you change this: >> >> cachedWriter = new PrintWriter(getOutputStream()); >> >>... to this: >> >> String encoding = result.getContentEncoding(); cachedWriter = new >>PrintWriter(new ObjectOutputSteam(getOutputStream(), encoding)); >> >>(or whatever you need to do to figure out the encoding) >> >>In SiteMesh we have to take into account that one request may have >>content written to both the outputstream and the writer, which is >>technically illegal in the Servlet API, but because SiteMesh munges >>multiple requests into one, we have to do some magic. However, I doubt >>this is necessary in OSCache, assuming the OSCache filter is applied >>before the SiteMesh filter. >> >>-joe >> >>On Thu, 31 Mar 2005 17:35:57 +1000, Scott Farquhar >> >> >> >>>On Wed, Mar 30, 2005 at 08:53:25PM -0800, Andres wrote: >>> >>> >>> >>> >>>>All getWriter() does is create a PrintWriter that wraps the output >>>>stream but the super.getWriter() should already do that. Unless I'm >>>>missing something this is convoluted and unnecessary object creation. >>>>If the new CacheHttpServletResponseWrapper's private writer field >>>>somehow adds behavior that the super's writer field does not, I do >>>>not see it. >>>> >>>> >>>> >>>> >>>This is not the case. The super method will not wrap around >>>getOutputStream(), hence the need for this method in the filter. >>> >>>I haven't had a chance to look at it, but I'd back the sitemesh way - >>>we've spent a long time trying to get the best method of working >>>inside of sitemesh, and it has been tested for i18n on almost all >>> >>> >containers. > > >>>Cheers, >>>Scott

Andres March - [31/Mar/05 02:39 PM ] Thanks for getting that in.

Lars Torunski - [31/Mar/05 03:15 PM ] Changes were done in the current CVS HEAD of OSCache 2.2 first and then backported to the 2.1.1 branch. Because a few issues were implemented for the release 2.2 before, I tried to minimize my modifications in the 2.1.1 branch.

Andres March - [12/Apr/05 11:31 AM ] I'm working on this. Something is not right.

Andres March - [12/Apr/05 11:34 AM ] duplicate of CACHE-38