google chrome cluster F on InfoWinow

[nmccready]

So it appears google chrome version 39 has fixed our CSS repaint woes.. However our repaint fix now breaks google maps on firefox and on the latest chrome.

Lovely art:

[nmccready]

Will we need something like this https://github.com/ded/bowser or is it provided in angular?

[nmccready]

Problem Child Code Lines here

[aesnyder]

are you thinking something like
chorme && chrome.version <= 39

[nmccready]

Changing

        showWindow: =>
          if @gWin?
            show = =>
              @scope.$evalAsync =>
                unless @gWin.isOpen()
                  maybeMarker = @getGmarker()
                  pos = @gWin.getPosition() if @gWin? and @gWin.getPosition?
                  pos = maybeMarker.getPosition() if maybeMarker
                  maybeAnchor = @getGmarker()
                  return unless pos
                  @gWin.open @mapCtrl, maybeAnchor
                  isOpen = @gWin.isOpen()
                  @scope.$evalAsync =>
                    ChromeFixes.maybeRepaint @gWin.content
                  @model.show = isOpen if @model.show != isOpen

to

        showWindow: =>
          if @gWin?
            show = =>
                unless @gWin.isOpen()
                  maybeMarker = @getGmarker()
                  pos = @gWin.getPosition() if @gWin? and @gWin.getPosition?
                  pos = maybeMarker.getPosition() if maybeMarker
                  maybeAnchor = @getGmarker()
                  return unless pos
                  @gWin.open @mapCtrl, maybeAnchor
                  isOpen = @gWin.isOpen()
                  @model.show = isOpen if @model.show != isOpen

works

[nmccready]

@aesnyder yup

[nmccready]

Where do you download older versions of chrome?

[aesnyder]

from my research you cant, client has 38 and one of my devs on my team as well...

[nmccready]

http://google-chrome.en.uptodown.com/mac/old

[nmccready]

not sure if it is safe

[aesnyder]

maaybe, brew cask?
https://github.com/caskroom/homebrew-cask

[aesnyder]

what is "safe" anyways?

[nmccready]

lol, yeah try out either one. Got the command to install an older chrome w/ cask? Vs point and click on the url fomr uptodown

[aesnyder]

I opened it from the dmg, and it showed up 39 when i clicked 'about'

[aesnyder]

looking for cask command now

[nmccready]

yeah dmg didnt work, but thats cayse chrome is using the same "Application Support Directory"

[nmccready]

IE cache

[aesnyder]

meh, cant get an oldversion from cask either

[nmccready]

firing up my ubuntu vm, maybe it is on an older

[nmccready]

ehh, im just gonna pull in bowser and make an angular service. We can unit test it and cross fingers and then drink heavily.

[aesnyder]

down

[aesnyder]

oh wait, i think browserstack would work

[aesnyder]

our subscription lapsed...

[nmccready]

Does your one developer still have 38? Can he verify that the InfoWindows still work with evalAsync without defer?

[nmccready]

Googleling
http://www.reddit.com/r/chrome/comments/1zvajo/how_can_i_have_multiple_isolated_chrome_browsers/
https://gist.github.com/uasi/882672

[nmccready]

Wow Browserstack rocks. Anyways 38 has same problem maybe GoogleMaps fixed this? Not sure.

[nmccready]

Same with 37 so looks like something we fixed or broke not sure. Can we do this without bowser?

[aesnyder]

yeah my dev saw that with evalAsync in 38

[aesnyder]

im not sure how its fixed, which makes me reticent to say that it is fixed

[nmccready]

Does evalAsync return a promise? How do we know when it is done? Both of these functions are ( or were) returning $evalAsync. To me it seems there is a problem with the creation not completing before showWindow actually happens.

[nmccready]

Looks like we started having repaint issue after, 27c2d27

I am gonna try removing the evalAsyncs and see if that issue works. I think the evalAsync in createGWin may be causing problems.

[aesnyder]

that might be the case, it doest return a promise but my understanding is that the callback will only execute after the dom has been manipulated by angular... in my mind kind of like animation frame

[aesnyder]

though as i read it this may not be correct, when called from a controller it happens before the view is rendered. i believe that the maybeRepaint should actually be triggered after render to get the height to render properly...

[nmccready]

@aesnyder has got chrome 38 to reproduce the original error reliably via browserstack, so this is most deff a chrome issue. Looks like we will pull in bowser to deal with it.