- From: thierry michel <tmichel@w3.org>
- Date: Wed, 12 Dec 2001 14:34:55 +0100
- To: "Chris Haynes" <chris@harvington.org.uk>, <www-forms@w3.org>
Chris, Thank you for your great review. The XForms WG will consider your input for the next XForms Working Draft version. ----- Original Message ----- From: "Chris Haynes" <chris@harvington.org.uk> To: <www-forms@w3.org> Sent: Tuesday, December 11, 2001 11:51 PM Subject: Feedback on Working Draft of 07 Dec > Dear XForms team, > > I've just read most of XForms 1.0 (WD 07 Dec 2001, ) as 'light' > evening reading. This is my first real entry into this topic. > > My initial comments and reactions may be of help. Even if I'm wrong, > they may indicate where other newbies might also have difficulties or > concerns. > > 2.1 Internationalisation > I find the claim that "Using XML 1.0 *ensures* ... internationalised." > a bit strong. > All XML seems to provide are: > - Internationalised characters sets, > - The ability to add 'xml:lang' attribute tags. > The implementer of XForms has *lots* to do herself in this area > (see 8.1 - Input - for example) > > > 2.1 Multiple device support > Surely you mean > "..re-present the user interaction on different devices." > not > "...re-purpose the user interaction to different devices." > The purpose is the same, the presentation is different. > > > 2.1 Declarative event handlers > I found the grammar clumsy here, you seem to be contrasting an > adverbial phrase ( "statically analyzable") to a noun phrase > ("imperative program code"). > > > 2.4 Providing instance data > I found the same identifier "as" occurring as an element name, an > attribute name and an attribute value confusing (but perhaps > unavoidable?). > > The example made me wonder if > > <payment as="credit" cc="" exp="" xmlns="..." /> > > would also be acceptable, > or alternatively if there was some permitted syntax such as > > <payment xmlns="..." > > <as default="credit" /> > <cc/> > <exp/> > </payment> > > > 3 Terminology - computed expression > 'relevant' and 'calculate' should be presented in a different font > (e.g. italic)., > or else the example put in brackets. > > > 6.4.2 Binding constraints last para > Spurious 'an' (or should be 'and' with comma removed?) > > 7.4.2.1 avg() > Picky mathematicians might say this should be called 'mean()'. > > > 7.4.3.2 now() > Surely the comment "(normalized to UTC)" belongs in the first > sentence - referring to the time, not in the second, referring to the > zone. > > > 8.2 textarea > Control of word wrapping > As far as I can tell, the combination of XForms 1.0 and CSS2 does not > support the functionality of the HTML textarea 'wrap' attribute. The > CSS 'overflow' does not really cover this. Although control of the > wrapping is a 'presentational' issue belonging in a style sheet, maybe > the XForms team has to lobby strongly to ensure this is supported in > CSS3. > > > 8.3 secret > A 'well known' browser has the useful behaviour that, if a page is > returned to using the browser's 'Back' function, the password field is > cleared. Would this be useful behaviour to mandate for the 'secret'? > Or maybe just say that is can *only* take input from the user(see also > 9.2 below) > > 8.6 range > "xsd:gDay" should not be in bold. > > 8.12.1 Common Attributes > The example states that the class attribute can have a space-separated > list of classes. > The comments below imply that only a single class may be used. > Which is it? > > 9.2 refresh > ...and other means of returning to a page, using Back or calling up > bookmarked pages. > As the developer of an application which uses long sequences of forms, > I find the > ability of the user to go back and re-present 'old' data at any time a > nightmare. > What I currently have to do is allocate a serial number to each page I > generate, and 'slap the user on the wrist' if she uses the back > button. > Can I put in a plea for some better, specified, behaviour with this > respect? > Would it be possible to mandate that forms with non-inline data should > always re-load that data each time the page is presented? > Leaving it to the user's manipulation of her cache control is not good > enough. > > 9.15 message > IMHO messages *must* be given an identifier, so that it is possible to > replace the content of the same modeless message panel, rather than > create a new one. I believe this to be a *vital* usability feature - I > could provide use cases if that would help. > > Review ended at section 10. > > There *is* someone out here reading this material :-) > > Hope this was of use, > > Chris Haynes > > > > > > > > > > > > > >
Received on Wednesday, 12 December 2001 08:34:58 UTC