======================================================================== http://mondrian.corp.google.com/file/10892071///depot/googleclient/gears/opensource/gears/desktop/drag_and_drop_utils_cr.cc?a=11 File //depot/googleclient/gears/opensource/gears/desktop/drag_and_drop_utils_cr.cc (snapshot 11) ------------------------------------ Line 26: #ifdef OFFICIAL_BUILD On 3:28 pm, nigeltao wrote: > Remove the #ifdef OFFICIAL_BUILD guard.
done. ======================================================================== http://mondrian.corp.google.com/file/10892071///depot/googleclient/gears/opensource/gears/desktop/drag_and_drop_utils_cr.cc?a=5 File //depot/googleclient/gears/opensource/gears/desktop/drag_and_drop_utils_cr.cc (snapshot 5) ------------------------------------ Line 241: // "Say" there's 1 drag file until we know more; extract the files. On 4:15 pm, nigeltao wrote: > On 2:58 pm, noel wrote: > > So why suggest that 0 more appropriate? In that case, we'd set g_drag_files > to > > 0, and return NULL to user javascript. > > Technical nit, but javascript doesn't have NULL -- it has null, and it has > undefined. meant context->SetReturnValue(JSPARAM_NULL, NULL), the js null. > But, what to return when dragging a directory -- whether to return 0, null, or > undefined -- is something I'll leave to when we do our cross-OS cross-browser > consistency pass. > > And consistency is why I want the Chrome code here to re-use > the same FileDragAndDropMetaData class that Firefox, IE and Safari > all use. That class as you've already mentioned is inconsistent and re-use is not my main concern. The code here runs in the context of the chromium browser, potentially within the chromium renderer, so I want to use solid apis. The code here checks all return values and errors. But there's one place I don't do that because of another api I have concerns about. > > What does > > event.dataTransfer.getData("Text"|"URL") return if we return NULL when files are being dragged? > On IE, getData returns null when there isn't any data of that flavor. exactly, and HTML5 is based on that model, and the GetDragData() api our designed to simulate event.dataTransfer.getData("Files"); so we'd return null if we do as you suggest, set g_drag_files = 0. And requests for .getData("Text"|"URL") would return null too, right? So what cursor does the page javascript show to the user in that case? Answ: never the right one. ======================================================================== http://mondrian.corp.google.com/file/10892071///depot/googleclient/gears/opensource/gears/desktop/drag_and_drop_utils_cr.cc?a=11 File //depot/googleclient/gears/opensource/gears/desktop/drag_and_drop_utils_cr.cc (snapshot 11) ------------------------------------ Line 375: g_identity = 0; On 4:22 pm, nigeltao wrote: > Any reason why we reset g_identity to zero? Does that mean that, if JS calls > getDragData twice in succession, we will hit the disk twice? yeah, was there reset g_identity in the improbable case of wrap, don't care about that no more. removed. ======================================================================== -- To respond, reply to this email or visit http://mondrian.corp.google.com/10892071
