- From: Philippe Le Hegaret <plh@w3.org>
- Date: 27 Mar 2002 17:02:10 -0500
- To: Steven Elliott <elliott@mail.cibertribe.pt>, John Keiser <jkeiser@iname.com>
- Cc: WWW DOM <www-dom@w3.org>
On Mon, 2002-02-11 at 15:14, Steven Elliott wrote: > I am not sure to what you are referring to as *differences between > implementations*. In any case I am not satisfied. > > Any means which permits the client to circumvent conditions placed upon the > form submit method invalidates the purpose of having an onsubmit method > (particullarly in light of the fact that it is used in 99% of cases as a > validation tool). To my mind this is (permits) a grave abuse of an implied > contract and the onsubmit method should be removed from the FORM object. At > the very least it should be boldly stated in any and all documentation that > the onsubmit event IS NOT GUARANTEED to be called by the submit method. > On Mon, 2002-02-11 at 04:57, John Keiser wrote: > It should be defined to not fire onSubmit. No JavaScript functions fire > events that I know of (at least in forms). onChange doesn't happen when > you change .value or .checked. It is not the user who is calling JS > .submit(). It is the page designer. We presume that the designer knows > enough about his app that when he calls submit() without validating he's > *deliberately* circumventing his stuff for whatever reason. This gives > the prog > > IE, NS4.x, NS6.x, and Konqueror work this way. What widely-used > implementations *do* fire onSubmit() that make it impossible to define > it this way in the DOM? As mentioned in the previous messages, there are differences between implementations regarding the HTMLFormElement.submit method. [[ In Spyglass Device Mosaic 3, the invocation of the submit() method causes the firing of the ONSUBMIT event prior to performing its intended semantics, and, if the event returns a value that is equivalent (coercable) to FALSE, then those intended semantics (form submission) are ignored. ]] http://www.w3.org/2001/12/DOM-Level-2-issues#i6 We cannot change this implementation even if NS and IE are consistent on this issue, therefore we reiterate our position: you cannot rely on having an event when invoking the submit() method. Philippe
Received on Wednesday, 27 March 2002 17:02:32 UTC