- From: John Kemp <john.kemp@nokia.com>
- Date: Mon, 23 Mar 2009 18:00:03 -0400
- To: www-tag@w3.org
Hello, I took an action to review the draft finding 'Usage Patterns for Client-side URL parameters' [1]. I consider this email sufficient to mark the action 'pending review' and will updated the action with a link to this email. I had various editorial comments, some of which related to use of terminology (such as 'web parts') and would like to know whether we are talking about URIs or URLs here (the document, however, seems quite clear that these are URLs). I noted also that the 'recommended best practices' section [2] of [1] is empty. I felt, however, that there were some potential best practices trying to reveal themselves from within the text: i) Web applications should use the fragment identifier to refer to client state-related information, as this helps links to remain "bookmarkable" and to make the browser back-button work reasonably. ii) Conversely, it sounds from the text as if the "naked hash-ref" is a potential anti-pattern. If we wish to publish this document in any form, we should certainly either fill in the empty sections or remove them. My raw comments are listed below. Regards, - johnk [1] http://www.w3.org/2001/tag/doc/hash-in-url [2] http://www.w3.org/2001/tag/doc/hash-in-url#id2261987 Full comments below ------------------- 1. s/URL/URI/g 2. What is a "Web part"? (Abstract) 3. Are # parameters the only (TAG-approved) way to pass parameters to the client? 4. Note: the abstract specifically mentions this as a *finding*. I would suggest that if we wish to publish this sooner rather than wait for Raman to make it a finding, we should make this text vaguer about the final output of the work. 5. 2.1.1 would like to see 'things to note' bulleted, or otherwise distinguished from one another. 6. 2.1.2 s/extrapollating/extrapolating/ 7. what, actually, is the common architectural model here? What is the difference between a JS app running inside HTML, vs. a video, vs. multiple iFrame proxies? 8. 2.3, 2.4 - 2.6 all seem rather similar. 9. 2.2 is not fully described - what exactly is the iFrame proxy doing to ensure the back button works? Is there a suggestion that everyone should do things this way? What situations does this work better in than some other method (simply updating the URL in the address bar with a # to maintain client state? 10. Is there actually a recommendation to be made here in using # parameters to carry client state for 'bookmarkability' - it felt like it from the text.... 11. Conversely, is "naked-hashref" an anti-pattern? 12. Remove or fill empty sections (ie. conclusions etc.) 13. Some references mentioned in text are missing, and at least one reference (to JSONP) is not referenced in the text (and in fact, what does JSONP have to do with #?) 14. In general, do we want to advise AGAINST breaking the same-domain policy by doing JSONP and/or iFrame proxy? Tough question (use-cases vs. security).... Should at least mention interaction here of such things allowing cross-site scripting or request forgery ("breaking" the browser same-domain policy), which is surely a security consideration.
Received on Monday, 23 March 2009 22:00:50 UTC