Hi Josh, Fixed all editorial comments below for this third set of comments. If satisfied with the corrections, please give us an OK for the DoC :)
On Thu, Jun 18, 2009 at 5:55 PM, timeless<timel...@gmail.com> wrote: > Date: Thu, Jun 18, 2009 at 6:48 PM > > 6:36 PM me: (e.g., a user agent could use an array, or an object, or a > hash map, etc.). > drop each "or" Fixed > 6:41 PM For each element in the elements list, if the element is one > of the following: > A preference element: > doesn't mention readonly (this might be ok, or maybe not...) Fixed. > 6:48 PM 1. For each file name in the default start files table (from > top to bottom) that has a media type that is supported by the user > agent: > 1. Let potential-start-file be the result of applying the rule for > finding a file within a widget package to file name. > 2. If potential-start-file is null or in error, ignore this file name > and move onto the next file name in the default start files table. > 3. If potential-start-file is a file, then: > 1. Let widget start file be the value of potential-start-file. > 2. Let start file content-type be the media type given in the media > type column of the default start files table. > 3. Terminate this algorithm and go to Step 9. > > I'm worried about the case where a package has two files: index.svg > and index.xhtml. index.svg is 0 bytes and index.xhtml is a well formed > xhtml file. The author developed this package using a user agent which > doesn't support image/svg+xml and things worked well. A user > unfortunately gets the widget and uses it with a user agent which does > support image/svg+xml. I'm fairly certain what happens is that this > process sends the user agent to step 9 with index.svg and the user > ends up unhappy. This would be the author's fault for including an empty file. The UA still does the right thing (it loads the index.svg file and finds it is not well formed.) Kind regards, Marcos -- Marcos Caceres http://datadriven.com.au