Describe/explain the bug
In case of using large datasets with ResponsiveBar the distance between axis y and the first bar is too large for some widths.
To Reproduce
I created a demo on Code Sandbox. As you resize the display area the distance between axis y and the first bar changes and it becomes too large for some sizes.
https://codesandbox.io/s/nivo-bar-tick-formatting-demo-j1zf9?file=/index.js
Expected behavior
The distance between axis y and the first bar doesn't become unproportionally large when resizing.
Screenshots

Same problem here, the bigger the data, the proportionally bigger the distance
I also observed that all of the bars will compress into one single spot right in the middle of the graph at around 406 bar elements

Anyone find a solution?
I ran through the related code a few weeks ago when I created this issue. I think it is caused by the calculations made by the d3 library but I'm not 100% sure.
Same issue here, it doesn't look good, and I couldn't find any solution apart from making excuses to complaining customers that it needs a fix from Nivo 馃槥
Question for people affected by this: have you considered using a Line graph?
From The example you provided:

source (forked codesandbox code from your example): https://codesandbox.io/s/nivo-bar-tick-formatting-demo-jso1b?file=/index.js:0-35617
I mention this because I remembered having had exactly the same issue in a couple of other libraries until it hit me: once you have a lot of data points, it makes more sense to just use a line graph to display your values. I think that's how I solved some graph issues here
@barnapisti1994 add the label:":bar_chart: bar" so appear as an open issue for that and let's hope it gets fixed soon
@drodmun I think only contributors can add labels to issues.
Any news for this issue ?
No progress here 馃槩
I think this is related to #840. If so there is a PR in the works to get this solved.
Looks to be fixed by #1282 and will be available in the next version. See here: https://codesandbox.io/s/nivo-forked-y50jx
Most helpful comment
Same problem here, the bigger the data, the proportionally bigger the distance