jp v2 - Call for Opinion

[sgreben]

Given that some time has passed since jp was released, it's a good moment to reflect on possible changes and improvements. Hence, the question for this issue is:

• What should change?
• What should definitely not change?

v2 will certainly include breaking changes- which is a chance to adapt the UX of the tool in a more drastic way than usual :) The timeline for this is roughly "until-end-of-year".

[gsauthof]

I'm interested in more support for visualizing distributions.

For example, adding a histogram plot type with vertical bars would be nice.

Adding boxplots would be great.

[Kabouik]

Late to the party, but I'm interested in plotting multiple columns (I use jp with csv files) against the same x variable, possibly with different colours. Also interested in datetime support for the x-axis.

[sgreben]

@Kabouik not late at all :) I only just got around to giving jp the rework it deserves this month :D

In particular with proper datetime axis support. My current WIP does this for the eth.csv example:

Re:multi-column plots - which plot type would you be interested in as a multi-series plot? bars or lines? Eventually both would be nice to have, but since I'm trying to get a first cut released quickly I'd rather do just one of those at first.

@gsauthof boxplots will also be included :)

ETA = 2-3 days

[mlaparie]

This look awesome. Glad to see that an update is coming soon with proper datetime support!

As you said, eventually both multi-series lines and multi-series bars would be nice, but I (with the other account above) was thinking about lines first because this is something I could readily use in phidget-tools, a program I worked on in the past few days. Plotting multiple variables of a CSV file on the same axes with different colours, instead of replicating plots as I currently do, would probably improve readability (as long as curves can be disabled, but that could be done outside jp I assume). Right now, plots are readable if I stack three, but four or five and they start to be too flat to give any sensible information.

I feel that the plot resolution either in scatter mode or in line mode is not optimal, compared for instance with plots from tools like bpytop, ytop or gtop. Are they using different braille symbols? bpyplot in particular has a very nice implementation as the symbols are big enough to be contrasty, but it just shows only vertical bars, not really line.

A few other features I was thinking about, not sorted by importance, and jp is meant to be a simple and frugal tool, so they may be out of scope:

• ability to set limits to x-axis and y-axis, optionally
• ability to add a label showing values of given data points; could be useful for static plots but also for dynamic ones on files that are being written, for instance one could show the value of the last row of a CSV variable, and run jp continuously so that it updates as the file is appended with new data
• grids maybe, but I like that jp looks clean and simple

Thanks again for the nice tool!

[dufferzafar]

@sgreben were your changes for v2 ever pushed? are they on some branch?

[Kabouik]

Here's another attempt at summoning @sgreben here!

What you've showed in 2021 looked great, and even though it was maybe preliminary or bugged, and may have stalled since then, we'dd really love to use it @sgreben. Hoping you can push it to some dev branch, with all usual disclaimers.

[aqilc]

I'd love to see something like stock market candlesticks or something similar! I've been trying to visualize stock market data and sometimes I just have a ton of JSON from various services, and while JP has helped me a lot, it would be much better if there were more types of graphs.