1398276 - "Disable cache" only to start

Status: NEW

(DevTools :: Netmonitor, defect, P3)

Description

[Peter Bengtsson [:peterbe]]

Attachment: Screen Shot 2017-09-08 at 2.56.04 PM.png

When you enabled "Disable cache" on the Network panel, it should disable any OLD cache but caching within the session. Like Chrome does it. 

Steps to reproduce:
1) Open Network panel and click on "Disable cache"
2) Open https://www.peterbe.com/css-blocking/bootstrap-lazyloadcsswithjs-vanilla.html

Expected results:
That the first load of bootstrap.min.css should be a 200 with an actual network fetch, but that the second request should be using caching.

Actual results:
There are two distinct 200 OK requests for bootstrap.min.css (same for that other jumbotron.css)

Also, I can confirm that the network traffic is real because I get these four lines in my server log when I refresh the page:

73.180.105.76 - - [08/Sep/2017:18:55:02 +0000] "GET /css-blocking/bootstrap.min.css HTTP/1.1" 200 18617 "https://www.peterbe.com/css-blocking/bootstrap-lazyloadcsswithjs-vanilla.html" "Mozilla/5.0 (Macintosh; Intel Mac OS X 10.12; rv:57.0) Gecko/20100101 Firefox/57.0"
73.180.105.76 - - [08/Sep/2017:18:55:02 +0000] "GET /css-blocking/bootstrap.min.css HTTP/1.1" 200 18617 "https://www.peterbe.com/css-blocking/bootstrap-lazyloadcsswithjs-vanilla.html" "Mozilla/5.0 (Macintosh; Intel Mac OS X 10.12; rv:57.0) Gecko/20100101 Firefox/57.0"

Comment 1

[Peter Bengtsson [:peterbe]]

The reason there are two requests at all for the same URL is to do with preloading followed by a piece of JS that changes the DOM and thus cause the browser to download it again. Yeah, that's nuts and a story for another bug but the point here is that the "Disable cache" should only be applicable for the stuff *before* the page is loaded.

Comment 2

[J. Ryan Stinnett [:jryans] (Use needinfo, replies may be slow)]

This might be good improvement, but I wonder how often this issue comes up for real pages...?  Hopefully it's not that common to request the same static files many times over on a real site...?  Perhaps <iframe>s change the story a little, I'm not sure.

Comment 3

[Firefox Bug Husbandry Bot]

https://wiki.mozilla.org/Bug_Triage/Projects/Bug_Handling/Bug_Husbandry#Move_fix-optionals

Comment 4

[Peter Bengtsson [:peterbe]]

(In reply to J. Ryan Stinnett [:jryans] from comment #2)
> This might be good improvement, but I wonder how often this issue comes up
> for real pages...?  Hopefully it's not that common to request the same
> static files many times over on a real site...?  Perhaps <iframe>s change
> the story a little, I'm not sure.

I don't think it's unrealistic at all. Preloading (see comment #1) is a good example of this. Either by things like `<link rel="preload">` or by doing things like `var img = new Image(); img.src=IMAGE_URL;`.

Comment 5

[:Harald Kirschner :digitarald]

Making this a parity bug as this becomes more impactful with the increase of caching options like preload.