series_annotations in bar plots in unusual positions for gr and pyplot backends

[JobJob]

Making a bar plot and trying to get labels on the top of each bar using series_annotations:

using Plots
gr()
# pyplot()
# plotly()
data = 1:3
labels = string.(data)
bar(data; series_annotations=labels, ylims=(0,5)) 

This works well with the Plotly backend, the labels go at the middle of the top of each bar.

With the GR and PyPlot backends the labels go in funny places

Plotly:

PyPlot:

GR:

Anyway clearly this package rocks in that I can just try another backend with a one line change and it works or looks better or is faster etc.

Cheers.

[mkborregaard]

This is because bar for gr and pyplot is handled by the recipe here: https://github.com/JuliaPlots/Plots.jl/blob/master/src/recipes.jl#L325-L378 , which draws the bars as shapes and thus changes the x and y values. For plotly the native bar implementation is used.

There could be two solutions:

1. to revert gr and pyplot to use the native bar implementation
2. perhaps to have the bar recipe create the bars in a second series with the @series macro, and return an invisible point series with the original x and y values?

[stugorf]

Has there been a solution developed for pyplot backend rendering of bar charts with series_annotations?

[mkborregaard]

no, unfortunately not. The best approach is to add a new series without markers with the same data, then then apply series_annotations to that.

[stugorf]

Thank you, I’ll try that! Regards,

…

-David

________________________________ From: Michael Krabbe Borregaard <notifications@github.com> Sent: Saturday, February 23, 2019 1:10:33 AM To: JuliaPlots/Plots.jl Cc: David Hughes; Comment Subject: Re: [JuliaPlots/Plots.jl] series_annotations in bar plots in unusual positions for gr and pyplot backends (#761) no, unfortunately not. The best approach is to add a new series without markers with the same data, then then apply series_annotations to that. — You are receiving this because you commented. Reply to this email directly, view it on GitHub<#761 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AFQ5eZXKDPPTcD9qUystn_NXxRGu8J98ks5vQQWJgaJpZM4M3lAk>.

[VarLad]

Any progress on this?

[VarLad]

no, unfortunately not. The best approach is to add a new series without markers with the same data, then then apply series_annotations to that.

I'm sorry, I didn't get that... can you please elaborate?