From: Dustin K. <dus...@us...> - 2006-06-27 02:51:25
|
On Fri, 2006-06-23 at 12:14 -0700, Martin Bligh wrote: > Bryce Harrington wrote: > > It would be nice to see autotest take an open view of this work, yet the > > reason we cc'd the autotest list was not to suggest that autotest adopt > > it, but primarily because we know folks at IBM and other companies were > > interested in mechanisms like this to enable easier exchanging of test > > data, and might like the opportunity to suggest improvements to the > > format. > > Nah, I'm not worried about it (or intending to do it). It could always > be post-processed by a external package if you really wanted to. Just > was outlining why I thought it was unnecessary. Historically, I've been partial to tests that output flat text. That said, I usually hack up some perl after the fact to process the data as I see fit. The flexibility I get out of perl/grep/awk/sed and flat text is something I would hate to give up. On the other hand, let me say that I do recognize the rich flavor of data encapsulated within and described by XML. And I certainly see where data marked up as such could provide a value-add to many (though not all) test results consumers. Unfortunately, I'm not previously familiar with the TRPI schema, so I have little perspective to give on the proposed changes. I do, however, have a few questions, which are along the lines of Martin's comment... - Where in the grand chain of test tools and infrastructure do you intend this functionality to live? - Is it to be implemented, as Martin suggests, as a post-processor, consuming established log formats (such as the LTP's) and producing fully decorated XML results? - Or do you suggest extending test suites to also support (hopefully not entirely replacing) directly producing TRPI results? - Is anything required of autotest in this space or was is this thread purely informational? :-Dustin p.s. What a garbled subject line :-) |