Bug 61452 – document.documentElement.cloneNode(true) causes resources to be downloaded again

• Status: NEW
• Component: Page Loading

[gildas]

Here are the steps to reproduce the problem :

1. Install Wireshark : http://www.wireshark.org/download.html
2. Open http://sergeydolya.livejournal.com/189087.html#cutid1 into a new tab
3. Start a capture with Wireshark and apply this filter:
http.request && http.request.method == "GET" && http.request.uri contains "web.jpg"
4. Type in URL input:
javascript:alert(document.documentElement.cloneNode(true)); void 42;

What is the expected result?
The browser shouldn't send again HTTP requests to get images content.

What happens instead?
Wireshark shows that images are downloaded again.

This issue is also reproducible with Chrome. With Firefox 4 and IE9, there are no extra requests.

[Alexey Proskuryakov]

These images are served with "Cache-Control:max-age=0, no-cache, no-store" header field, so obviously they can't be cached. 

Cloning an existing element can probably use the same image resource indeed - no need to request it from the network layer.

[malch]

I have recently encountered this bug and offer a few additional observations:

1. There is no need to install a packet sniffer to observe the problem. One can view the unnecessary network activity from the Network Panel of the Chrome Developer Tools frame, for example. Or monitor the access log on the server hosting the image of course.

2. Although the problem seems to generally arise on images with no-cache, it can also be seen with regular static images that are 404 Not Found.

3. In my case, cloneNode() was being called from 500ms timer. Thus the level of unnecessary HTTP requests being generated was quite significant and multiple users caused a serious load problem for the server hosting the image.

[Alexey Proskuryakov]

*** Bug 264657 has been marked as a duplicate of this bug. ***

[Radar WebKit Bug Importer]

<rdar://problem/121406746>