I do not think always passing UTF-8 as character encoding type is right behavior. As far as I can tell, there is no such requirement to mandate(or even recommend) UTF-8 as character encoding type in the HTML4 specification (and relevant RFCs).
A while ago, I submitted a patch that enables the encoder to take character encoding name as second parameter. I see no reason why the character encoding is done with UTF-8. Using UTF-8 as character encoding type works only if the html documents sent from the application are encoded in UTF-8, but many i18n applications use 'native' character encodings (like Shift_JIS, Big5, EUC-KR etc) instead, for variety of reasons. The issue was recently discussed on the mailing list of Japanese struts user's. And we found that, for example, html:link tag's parameter populating functionality has been unusable for many of us because it ignores the response's character encoding and uses UTF-8 to get the bytes for url encoding. > --- Robert Leland <[EMAIL PROTECTED]> wrote: > > Pierre Maury wrote: > > The patch would need to be usable with different encoding types. > > Why? The Java 1.3 version doesn't accept an encoding type and we always > pass UTF-8 to the 1.4 version. > > > This memory storage seems like a small price to pay for the improvement, > > but I would like to hear from other commiters. > > I'm not a big fan of this because there are much bigger performance > optimizations to be had in other layers of a Struts app but I won't stop > anyone from applying a clear and *tested* optimization. > > David > > > > > -Rob > > > > >Hi, > > > > > >I think there is a performance issue with > > >org.apache.struts.taglib.TagUtils.encodeURL(String). > > > > > >Attached to this mail, four java classes that demonstrate and solve the > > >problem. You may run EncoderTest to compare the timings of the original > > >struts code with a faster encoder. > > > > > >On my computer encodeURL takes about 0.2ms. This does not look like a > > big > > >issue, but encodeURL is called by computeUrl for each parameter name > > and > > >each parameter value. The web site I'm working on displays products > > lists. > > >There an average 3 links per product and 20 products per page + 20 > > other > > >links. Each of this links takes four parameters. > > > > > >Total encodeURL time per page amounts to : (3*20+20)*(4*2)*0.2 = 128ms > > ! > > >At the same time, the struts action + the JSP page takes about 200ms. > > > > > >The attached encoder is at least 50 times faster. By using this encoder > > on > > >my application, the troughput was greatly improved. > > > > > >If you are interrested, I can submit a patch > > > > > > > > >Pierre Maury > > > > > > > > > >------------------------------------------------------------------------ > > > > > >--------------------------------------------------------------------- > > >To unsubscribe, e-mail: [EMAIL PROTECTED] > > >For additional commands, e-mail: [EMAIL PROTECTED] > > > > > > > > > > __________________________________ > Do you Yahoo!? > Free Pop-Up Blocker - Get it now > http://companion.yahoo.com/ > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] ------------------- Yasuhiko Sakakibara [EMAIL PROTECTED] [EMAIL PROTECTED] --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
