IFrameNavigator’s start inconsistencies

[JayPanoz]

So I’m currently debugging that but once again, it’s unclear when I’ll be able to commit it, and it also depends on how we’ll be handling the merges between Bokbasen and Jellybooks. So filling the issue to let people know.

There’s quite a nasty inconsistency related to iframe’s loading across browsers. And we’ve been impacted in public beta – sigh.

By default, it doesn’t have a src in the codebase, which means (non-chromium) MSEdge and Firefox are forcing an about:blank as a src when the IFrameNavigator is started.

In other words, what should be:

1. Load manifest
2. handle iframe load

is the following in those browsers:

1. handle iframe load with about:blank
2. load manifest
3. handle iframe load with spine item

The only thing that is preventing the reader to fail in MS Edge and Firefox on subsequent uses is that the manifest is requested every time a chapter is loaded.

When I tried to minimise the number of such requests because I saw quite mind-blowing stuff (e.g. 8 requests of the same manifest to the server within a few seconds) i.e. request it on start then use the one in memory, all hell broke lose, and I found myself in a never-ending loop of blank pages – which explained the bug some users had encountered in public beta.

In other words, you’d better protect yourself against about:blank as the currentLocation in the handleIframeLoad method – it’s still unclear to me but apparently, Chrome and Safari are supposed to do the same in the foreseeable future.