- From: Klotz, Leigh <Leigh.Klotz@pahv.xerox.com>
- Date: Fri, 9 May 2003 16:01:36 -0700
- To: "'Dueck, Brian'" <bdueck@metasolv.com>, "'www-forms@w3.org'" <www-forms@w3.org>
- Message-ID: <51B8ABCE456FD111899900805F6FD6EE12EE00DC@mercury.ADOC.xerox.com>
I agree with Micah that undo/redo of user interaction should be possible in user agents without any blessing or encouragement from a W3C spec. Just as a textarea widget in a browser can interpret Ctrl-Z and undo and Ctrl-Y as redo without spec guidance, a menu item in a desktop browser could offer Undo/Redo support across widgets -- undo typing, followed by undo checking box, followed by undo sliding slider. For your point 1, don't forget voice and other input modalities. If undo and redo are to be initiated from events inside the form, however, there would need to be some additions to the specification. Your second and fourth points are also good areas to look for implementation and use experience before writing them down in a spec, but I think implementations can make significant progress without dealing with any hard issues involving side effects. Sidebar on reset: You might note that XForms does not have a <reset> widget. One of the thoughts behind de-emphasizing <reset> is that users would better be served by the user agents offering undo/redo support as part of their user interface. There were other thoughts too -- form designers put in reset because it's there, not because they think people want it, and although they style the submit button they leave the reset button a big gray box that really wants to be pushed. And there's no undo ;-). For my pet-peeve example of this practice, see http://mapsonus.switchboard.com/bin/maps-mapspec/ <http://mapsonus.switchboard.com/bin/maps-mapspec/> The "Draw Map" button is unfamiliar looking, out of style with the rest of the page, and located under the input labels, whereas the "Clear" button looks like a button to push, and is located right under the input fields you've just typed in. Leigh. -----Original Message----- From: Dueck, Brian [mailto:bdueck@metasolv.com] Sent: Tuesday, May 06, 2003 9:52 AM To: 'Micah Dubinko'; 'www-forms@w3.org' Subject: RE: XForms and Undo Thanks for the response Micah. I think what's missing from the spec somewhere is to explicitly describe how XForms 1.0 supports an Undo scenario. You're right I'm sure that XForms 1.0 probably has enough to support basic Undo, but I think the feature may be important (and complex) enough to warrant an explicit mention (Undo may not be alone in this). Making the behavior clear and explicit will I think encourage XForms implementations to support this capability and that (I believe) is good for everyone concerned. Here are some of the questions that may be helpful to answer for XForms implementors and authors interested in Undo: * Explain what the expected functional behavior is of Undo in XForms (this may be painfully obvious to some, but I put it in the same category as specifying things like what the visual effect of the "relevant" model item attribute has on UI controls - it's better to be explicit so that the form author can be assured of providing the end user the same experience across XForms implementations). * Is there a standard way for a form author to hint to an implementation that Undo support is desireable and practical for a given form? * What is the implication of a series of Undo operations on the model (this may be no different than a regular edit, but if so it still makes sense to clarify). * Identify the features of XForms 1.0 will be problamatic or nonsensical to accomodate Undo. Submit may be an example of this, but I suspect other events may be difficult or perhaps impossible to reliably undo (e.g. what is the undo of an event that triggers a message to popup? External and date based functions may also have side effects that are not possible to undo.). * A close functional cousin to Undo is of course Redo (i.e. undo my undo, or alternatively repeat my last command). It would be useful to cover these topics as well and explain the expected behavior from a user perspective. Does this make sense? Brian. -----Original Message----- From: Micah Dubinko [mailto:MDubinko@cardiff.com] Sent: Tuesday, May 06, 2003 3:56 PM To: 'Dueck, Brian'; 'www-forms@w3.org' Subject: RE: XForms and Undo Hi Brian, Yes, we have discussed undo, and we think it's a good idea. An implementation would certainly do well to support an undo feature (and indeed, it's fully possible even with XForms 1.0). What would you suggest putting in the specification to encourage this feature, or make it more explicit? Thanks, .micah -----Original Message----- From: Dueck, Brian [mailto:bdueck@metasolv.com] Sent: Monday, May 05, 2003 11:31 PM To: 'www-forms@w3.org' Subject: XForms and Undo Hello; Wondering if anyone has given thought on how XForms addresses the common user need to "undo" a series of edits in historical sequence. For example, through an XForm, a user is editing some important information, realizes a mistake has been made, and needs to "undo" several of these edits. The result of each undo is the changes made for that edit to the XForms models are reversed. Perhaps some actions denote a commit point that cannot be undone and clears all prior edits - for example a submit. I can imagine this would be a very powerful feature for an XForms implementation to provide to its users. This would be especially nice if the Undo capability was done in such a way as to be zero effort for the XForm author. I believe end-users have come to expect undo from their commercial applications (e.g. office productivity apps, etc.), so why not address this requirment as part of the XForms spec? That way the behavior associated with an Undo is consistent from an end-user perspective as well as an XForms author perspective. Or if this does not need to be part of the spec perhaps the spec can at least explicitly say its up to implementations to provide this feature or not. Has this been addressed or discussed explicitly by the XForms spec groups? How about the XForms implementation community? Anyone stepping up to the plate to provide this feature? Thanks in advance, Brian.
Received on Friday, 9 May 2003 19:01:37 UTC