Invert Stacking

[mbarkhau]

In the current rendering, when lower fee transactions are added to the mempool, higher fee transactions are pushed up, which doesn't reflect anything about reality. If the vertical axis were inverted and rendered from top to bottom, then adding higher fee transactions pushes lower feed transactions deeper into the mempool. (see even the language works better).

[mbarkhau]

I guess it's really the stacking that has to change. The lower fee transactions should be stacked on top of the higher fee transactions. I know I may not be making myself clear, I'll see if I can adapt mempool.js.

[mbarkhau]

Here I've simply reversed the order of stacking done in convertData. I think this makes some things easier to read.

1. High fee transactions visually fluctuate less, because they are not moved by meaningless transactions with lower fees.
2. It is easier to draw a line into the future, and have a guess at when transactions with a given fee will be processed (when will the drawn line cut y = 0).

[mbarkhau]

This is maybe better for comparison. Now that I look at it, I honestly don't know if it makes so much of a difference. The chart looks a lot more busy with the inverted stacking and I'm not sure it conveys very much more information.

[mbarkhau]

Ok, they're both crap, I think a 3d plot would be better.

[mbarkhau]

This WIP pic still looks like crap, but it uses the same chunking of the stacked charts. I think it will look better if the sat/byte axis has a finer resolution.

[jhoenicke]

I'm wondering if a logarithmic scale for the size graph makes sense, stacked in reverse order. This may make it easier to see which fee levels confirm next.

[Giszmo]

The inverse chart is uglier but it's more stable when hiding low fee levels or in other words, the user will not need to hide fee levels if he sees the result ignoring a color, instantly.

As a default, I'm against the implementation of this "feature".

[murchandamus]

I was interested in this feature to more easily see what level of fees will still be part of the next block. I do agree that the inverted graph looks very busy, especially since new blocks will whack out a piece from the bottom. Better would perhaps be to add an option to only show the first n times 4,000,000 weight units of the mempool.