If one visits any page, and then loads a frameset page, the frameset is drawn before the content of the frameset is loaded. The orignal page is then loaded into the framset for a brief instant. Aesthetically unpleasing. REPRODUCE: 1. Load any URL into apprunner. 2. Manually enter a URL of a known frameset into the location box and hit ENTER. 3. Watch as the frameset is drawn and the contents of the URL #1 are drawn into each frame for a brief instant. USING: Nightly build (4-10-99 Win32).

I went to sample 6 and then to sample 9 which is a frameset document. I didn't see the problem on 4/10 apprunner or 4/10 viewer. Reassigning to David to take a further look.

The bug description isn't quite accurate. Chris, in both viewer and apprunner (I see it in both), there's a noticeable delay between the time the initial splitter bars are drawn for the new frameset and the time that the frames are actually filled in with data, i.e., the splitters appear, but the subframes' contents aren't loaded yet, and the old page isn't being blown away by an EraseBackground call in the subframes until the body tag is being encountered in the subframes' subdocuments. Maybe what needs to happen is that you need to go ahead and wipe a new webshell's view and fill it with a background color or something when it is first constructed. I suspect that as performance improves, this bug will become a non-issue anyway. For me, sample #9 takes a really really long time to load, and I think that's really the only reason I have time to see all of this stuff. Anyway, I see it in viewer and in apprunner, so I'm bouncing it back to you, Chris. :)

Reassigning to Nisheeth, since it sounds more like a web shell problem.

I am a little mystified here at the inability of others to see this. I am trying it with the 4-10-99 build as I write this. It is most defnitely reloading the contents of the original page into each frame cell BEFORE loading the actual content of the frames.

Screenshot captured. Take a look at http://holly.colostate.edu/~jerbaker/Mozilla/mozilla_bug.gif

Confirming. It doesn't seem to clear the page before relodng the next one. Checked using Win98.

I'm pretty sure this is related... when using the target=_top it first copies the original screen into each frame, then it starts over. (See http://www.novagate.com/~jsteenhagen/Mozilla/4947.html ) (Windows 98)

Accepting bug. Setting component to Embedding APIs and target milestone to M6...

Moving bug to M8...

Putting on [Perf] radar

Distibuting bugs over milestones... This one goes to M10.

*** This bug has been marked as a duplicate of 5849 ***

verified dup

Mass removing self from CC list.