I've tested with the released Safari 3.0.4 and the r27375 webkit and whenever I load a page with Flash elements that have been previously downloaded, they are downloaded again. You can watch them re-download in the Activity window and I've confirmed the download requests on the server logs. If I run r27375 under Tiger with Safari 2 installed or run Safari 2 straight, the problem does not exist. Nor does it exist when running Firefox. So, it seems the problem is limited to either Safari 3 or Safari 3/Leopard. I don't currently have a Tiger system with the Safari 3 beta installed to check and see if it's Leopard specific. Note on the test link: The site is in development and the index page is blocked unless you use the full link with index.php at the end. The home page links will go to the blocked page, so if you want to go back to the home page, you'll need to use the link provided. The home page has 2 flash files you can view the download on, but every page has the flash header which best visually illustrates the issue. It will say loading until the whole flash file is loaded and should immediately jump to a picture if the flash file was being cached.
Anyone ever checked to see if they're seeing the same? In the previous comment, you can ignore everything from "Note on the test link" down since the site is now live and the home page no longer has 2 flash files. There's just the one as the header on every page now.
We're seeing the same behavior at http://www.picnik.com. We have a 1.4MB SWF file that we'd really like to be cached, but Safari reloads it every time. Repros on both Windows and Mac. Other browsers load from the cache appropriately. Hit this url to repro: http://www.picnik.com/app
I can confirm this bug. Regressions are P1.
<rdar://problem/5759372>
This bug is not in WebKit, but in a framework that WebKit uses for HTTP access. Marking as invalid to indicate it is an issue outside of WebKit.
(In reply to comment #6) > This bug is not in WebKit, but in a framework that WebKit uses for HTTP access. > Marking as invalid to indicate it is an issue outside of WebKit. > Regardless of its source, this bug is still affecting Safari users. Can you give us the name of the framework, or a pointer to a description of the bug in said framework, so we can take it up with those developers?
Apparently this is fixed in MacOS 10.5.3.