Minor Steps to reproduce:
- Install recommended plugins
- Create a freestyle job named 'foo'
- radiobutton 'Git'
- Click 'Advanced…' to show Advanced Project Options
- Check 'This build is parameterized'
- Add two boolean parameters (no need to specify name or anything)
- Check 'Retry Count'
Expected result: Retry Count gets checked
Actual result: It does get checked, but the page scrolls down.
[JENKINS-33891] Unexpected scrolling on job configuration page
gus reiber
added a comment - Can reproduce in Firefox... checking against 2.0 source branch. gus reiber
added a comment - bug is present in 2.0 as well.... digging around. gus reiber
added a comment - http://jenkins.beedemo.net/api-team/job/sandbox/job/hello/configure
This behavior exists in v1. The above steps can be done against the beedemo URL to get the same jumping behavior with FF.
scrollIntoView at ln 1367 of hudson-behavior.js does not correctly calculate the page size after things have been opened, and thus scrolls the window to the wrong spot. The easy fix is just to remove this behavior. Perhaps a better fix would be to fix the math, but I am not convinced that really is the right behavior and I am not sure how best to jump through those hoops, in prototpye.js with who knows how many page shape changing events may happen on the page.
gus reiber
added a comment - http://jenkins.beedemo.net/api-team/job/sandbox/job/hello/configure
This behavior exists in v1. The above steps can be done against the beedemo URL to get the same jumping behavior with FF.
scrollIntoView at ln 1367 of hudson-behavior.js does not correctly calculate the page size after things have been opened, and thus scrolls the window to the wrong spot. The easy fix is just to remove this behavior. Perhaps a better fix would be to fix the math, but I am not convinced that really is the right behavior and I am not sure how best to jump through those hoops, in prototpye.js with who knows how many page shape changing events may happen on the page. gus reiber
added a comment - https://github.com/jenkinsci/jenkins/pull/2193
I have offered a PR that fixes the issue specifically for v2, as I am not sure if it doesn't mostly work enough in v1 that people like it, despite being broken in some cases. I don't think it is worth fixing, but others might think so. danielbeck, do as you will....
gus reiber
added a comment - https://github.com/jenkinsci/jenkins/pull/2193
I have offered a PR that fixes the issue specifically for v2, as I am not sure if it doesn't mostly work enough in v1 that people like it, despite being broken in some cases. I don't think it is worth fixing, but others might think so. danielbeck , do as you will.... gusreiber It's not useful to link to private Jenkins instances. Not even I know how to access that one (and for the purposes of this issue tracker, I don't want to know).
Reproduced on pristine 1.642.2 with same steps.
Not a regression, and apparently rare serious enough for users not to complain in numbers, so let's leave this until after 2.0.
Tiago Lopes
added a comment - Still happening on 2.10, checking an optionalBlock with the property ' inline="true" ' is causing the page to scroll to the bottom on Firefox
Tiago Lopes
added a comment - Still happening on 2.10, checking an optionalBlock with the property ' inline="true" ' is causing the page to scroll to the bottom on Firefox 
James Holderness
added a comment - I encountered a similar problem on the management Configure System page, while clicking on a checkbox for an optional block. I eventually tracked down the issue to a bug in the scrollIntoView function in hudson-behaviour.js (see here). This seems a generic enough function that I could imagine it causing problems in other places too.
The issue is that it's calculating the distance to move relative to the current scroll position, but it's obtaining the current scroll position from document.body.scrollTop which is always zero. It should be using document.documentElement.scrollTop (or possibly a combination of the two for backwards compatibility). Chrome used to support body.scrollTop, but that was a bug, and once that was fixed, anything that relied on the bug stopping working. You can read more about it in the Chromium issue tracker here and here.
This can be a difficult issue to reproduce, because sometimes it appears to work. But that's just because of another bug in the updateOptionalBlock function, which often returns a lot of NaN coordinates for the block's visible region. The result being that it just doesn't scroll at all (which is at least better that jumping to the bottom of the page).
James Holderness
added a comment - I encountered a similar problem on the management Configure System page, while clicking on a checkbox for an optional block. I eventually tracked down the issue to a bug in the scrollIntoView function in hudson-behaviour.js (see here ). This seems a generic enough function that I could imagine it causing problems in other places too.
The issue is that it's calculating the distance to move relative to the current scroll position, but it's obtaining the current scroll position from document.body.scrollTop which is always zero. It should be using document.documentElement.scrollTop (or possibly a combination of the two for backwards compatibility). Chrome used to support body.scrollTop , but that was a bug, and once that was fixed, anything that relied on the bug stopping working. You can read more about it in the Chromium issue tracker here and here .
This can be a difficult issue to reproduce, because sometimes it appears to work. But that's just because of another bug in the updateOptionalBlock function, which often returns a lot of NaN coordinates for the block's visible region. The result being that it just doesn't scroll at all (which is at least better that jumping to the bottom of the page). 
So far cannot reproduce with OSX Chrome Version 49.0.2623.87 (64-bit)