Positioning off on mobile devices

[henrikingo]

On a motorola android at least, the current slide has its center in the top left corner of the viewport. See screenshot.

[henrikingo]

[janishutz]

Ohh jeez. Do you have any idea what could cause it?

[janishutz]

Fun fact is, I actually think that only parts of impress.js might be supported in that browser. Because for me it works perfectly on mobile

[henrikingo]

This seems to be an issue (still) on mobile Chrome and Opera, which is a Chromium derivative. But on mobile Firefox this works great. So I'm going to assume this is actually a bug in Chrome-family browsers.

[gborgonovo]

Since Chrome is the most used browser, is there any solution or workaround to this issue?

[janishutz]

@gborgonovo I am still looking into a full rewrite when I get the time to do it. Cannot give an ETA, but it's not that far away. I know most of the maths now and I have parts of it figured out, but some things still need figuring out. And I need to draw up the full API as well to make it intuitive

[gborgonovo]

@janishutz great! Thank you very much!

[henrikingo]

@janishutz Can you please stop talking about a full rewrite on every issue. If you have code to share, then share it. If not, please don't do this, because it kills every conversation where others could discuss fixing the actual bugs in the current code (or in this case, probably the browser).

Also, even if you did do a full rewrite, there's no reason to expect that I would approve of that code to replace impress.js. For most users it would make sense for a full rewrite to be a new repository. So, for this and other issues, let's keep this repository focused on the current impress.js code, and people who want to use and improve it.

[janishutz]

Sorry about that.

The point was to mention there is a possible solution coming eventually, but I should have mentioned that we could try to fix it. However, I believe the issue to be a fundamental issue in the way impress renders that might clash with a Blink bug. If it really is, then fixing it without some larger amount of refactoring is unlikely going to be possible.
I will have some code to share in a few weeks, but I don't want to rush it and I want to keep as much of the API as possible the same (thus needing more attention), whilst enabling a "free-roam" plugin to work out-of-the box, as it will just tell core to move to position (x, y, z)