I've discovered a possible work-round for the different sample element names - just assume that all the immediate children of /testelement are samples. Not sure if this would work with all possible output files, but the approach looks promising so I'm no longer proposing changing the names.
I still think it's worth changing the encoding of strings to use the Xstream default. See the separate e-mail about that. S. On 7/2/05, sebb <[EMAIL PROTECTED]> wrote: > I've just discovered that the latest version of xstream (1.1.2) does > the required escaping for string attributes. > > I'd still like to fix the sample result element names. > > S. > On 7/2/05, sebb <[EMAIL PROTECTED]> wrote: > > I've been trying to update the jmeter-results*.xsl stylesheets in the > > extras directory to work with the new test result outout format. > > > > This is proving rather messy, so I'm wondering if the output format > > could be changed? > > > > == > > > > One problem is the escaping of attribute string values. The original > > implementation used standard xml constructs - < " etc, and did > > not escape spaces, whereas the current implementation uses > > URLEncoder.encode(). > > > > This means that the xsl stylesheet now has to decode (de-escape) the > > strings before display, whereas previously, the browser would do the > > decoding automatically. > > > > The escaping needed is not very complicated - " -> " etc > > I've not yet found a way to access the code that currently does it, > > but it can be recreated fairly easily. > > > > == > > > > Another problem is that there are now two different sample elements - > > <sample> and <httpSample>, whereas previously there was only > > <sampleResult>. > > > > This means that walking the tree to find the samples is quite a bit > > more complicated, as one has to check for two different types of > > element. If we ever introduce more sample result classes, the problem > > will just get worse. > > > > So I'd like to suggest that we revert to using just <sampleResult>. > > This would mean a bit more work when unmarshalling the xml, as the > > appropriate Result class would have to be determined prior to creating > > it. > > > > == > > > > Thoughts? > > > > Sebastian > > > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
