- From: Curt Arnold <carnold@houston.rr.com>
- Date: Wed, 7 Mar 2007 12:46:02 -0600
- To: Dominique Hazael-Massieux <dom@w3.org>
- Cc: public-mwts@w3.org
On Mar 7, 2007, at 2:51 AM, Dominique Hazael-Massieux wrote: > >> The self-hosted productions were never endorsed >> by the DOM WG and the results were never fully compared to those run >> using JSUnit. > > Could you comment a bit more on this? Is there any reason to think the > self-hosted version may be less reliable than the JSUnit-based one, or > is it simply that nobody had the time to make a proper comparison > between these two test harness? > > Thanks again, > > Dom The self-hosted production was done late in the lifetime on the DOM Test Suite effort to try to address running the test on resource limited browsers, but it was more of a experiment than a WG product and did not have the same level of review. The test documents (hc_staff.html and the like) are renamed and a few elements added to produce the self-hosted file. Most tests are not sensitive to those modifications, but a few may be. I think it is likely that the the DOM L1 Core, L2 Core and L2 HTML have had all those issues addressed for HTML (and maybe XHTML) in the CVS, but other modules may not have been reviewed. It might be interesting to compare the currently generated self- hosted tests with the snapshot in the Webkit SVN. I helped the project incorporate the self-hosted tests into their build system about 2 years ago, but have not followed any subsequent modifications. The WebKit variant of the self-hosted tests is located in the LayoutTests/dom directory of their source tree and can be checked out using: svn checkout svn://anonsvn.opensource.apple.com/svn/webkit/trunk/ LayoutTests/dom WebKitTS There are several places where the WebKit developers (appropriately) chose compatibility with IE and Mozilla over compliance with the recommendation (or at least the test suite interpretation of the recommendation) and those decisions may be reflected in either the expected results for the tests (the corresponding .txt file) or the tests may have been modified. There have been a few issues raised on the www-dom-ts@w3.org or www- dom@w3.org mailing list over the last many months that appear worthy of at least discussion. I have had good intentions to address those, but the combination of little free time and no clear process since the DOM WG expired has resulted in it never getting to the top of my list. p.s. I have subscribed to this list and no longer need to be cc'd.
Received on Wednesday, 7 March 2007 18:47:09 UTC