IE8: selection methods not available.

[krusynth]

In IE8, things like document.createRange and window.getSelection don't exist. Per the discussion over on Annotator openannotation/annotator#412 (comment) we're suggesting users use the Rangy plugin to fill these gaps - but that doesn't polyfill window or document, it just provides a standard interface instead a la jQuery.

For instances like this and this I think it would make sense to test if rangy is being used and if not, default to the current methods. E.g.:

win = rangy ? Util.getGlobal()
sel = win.getSelection() # Get the browser selection object

Relates to openannotation/annotator#428

[krusynth]

One more instance.

[csillag]

@nickstenning @tilgovi What do you think?

[tilgovi]

Seems fine to me.

[tilgovi]

But lets call the variable something other than win... not sure what... maybe ctx?

[krusynth]

That's certainly doable - but it is a replacement for window in most of these cases.

[tilgovi]

Yeah it's not a big deal either way. These functions would typically be
on the window, of course. But here they're not, so the win variable name
made me do a little wtf.
On Sep 19, 2014 2:53 PM, "Bill Hunt" notifications@github.com wrote:

That's certainly doable - but it is a replacement for window in most of
these cases.

—
Reply to this email directly or view it on GitHub
#4 (comment)
.

[nickstenning]

I understand the need for a replacement for window.getSelection() but I don't understand why this issue is on this repository. The classes in this repository operate on selection objects, but they don't (I don't think) need to know where they came from, do they?

[krusynth]

@nickstenning This repo calls out to window.geSelection directly, so yes, some sort of replacement is needed. (See my patches above.)

[tilgovi]

That's maybe a good thing for us to work toward changing.
On Sep 20, 2014 6:43 AM, "Bill Hunt" notifications@github.com wrote:

@nickstenning https://github.com/nickstenning This repo calls out to
window.geSelection directly, so yes, some sort of replacement is needed.
(See my patches above.)

—
Reply to this email directly or view it on GitHub
#4 (comment)
.

[nickstenning]

Am I being an idiot? The only reference I see to getSelection in this repository is in the function Util.readRangeViaSelection, which in turn is only used in the tests.

[krusynth]

There's also createRange. opengovfoundation@7e3dd65

[nickstenning]

So, it doesn't call getSelection in a way that matters. The tests can probably be modified.

This repository does however, call document.createRange in NormedRange#toRange. This is only used in one place in Annotator and can easily be removed.

[krusynth]

@nickstenning you clearly understand the problem better than I do, feel free to have at it and I can amend my PR to revert my proposed changes here.

[nickstenning]

Okay, brilliant. Thank you, @krues8dr, for identifying this coupling. I've submitted #8 here to refactor the test, and have committed openannotation/annotator@e28f193 to remove the need for the NormalizedRange#toRange() method. With #8 merged, we can remove this method from this repository and with it the last explicit reference to the W3C Range APIs, leaving rangy detection to be done elsewhere.

[krusynth]

@nickstenning Great, so that's half of it - but there's still getSelection to deal with as well. And also this repo will need to be refactored for that change.

[nickstenning]

There are no uses of getSelection in this repository that will remain after #8 is merged and NormalizedRange#toRange() is removed.

[krusynth]

Great, so I've merged that in to our fork and tests are still working. I'm at a conference atm so I can't dig too much on this right now, but it looks good at first glance.

[tilgovi]

It sounds like this was resolved. Please re-open if that isn't so.