Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • 12:03
    coding-krtek commented #21369
  • 09:50

    timhoffm on v3.5.x

    Backport PR #21864: Delete "Loa… Merge pull request #21865 from … (compare)

  • 09:50
    timhoffm closed #21865
  • 08:51
    meeseeksdev[bot] labeled #21865
  • 08:51
    meeseeksdev[bot] milestoned #21865
  • 08:51
    meeseeksmachine opened #21865
  • 08:51

    jklymak on main

    Delete "Load converter" example… Merge pull request #21864 from … (compare)

  • 08:50
    jklymak closed #21864
  • 08:17
    noonchen commented #11607
  • 02:01
    brobr commented #21718
  • 01:56
    timhoffm commented #21799
  • 01:46
    janash commented #21369
  • 01:44
    timhoffm labeled #12915
  • 01:44
    timhoffm closed #12915
  • 01:44
    timhoffm commented #12915
  • 01:22
    timhoffm labeled #21864
  • 01:22
    timhoffm milestoned #21864
  • 01:21
    timhoffm opened #21864
  • 01:10
    timhoffm edited #21863
  • 01:09
    timhoffm labeled #21863
Nathan Goldbaum
@ngoldbaum
beginning with the black navbar
and now this apparently :)
Thomas A Caswell
@tacaswell
unfortunately this frog is noticing being boiled
otoh, there is as a story of one of the big retailers (I think walmart?) spending lots of money on a complete overhaul of their site
and then trashing almost immediately after launch because it crashed sales
because people knew how to navigate their old site and when the site changed abandoned purchases despite the new site being 'better' in many ways
also why amazon and google have never done one-shot make overs
Aleksey Bilogur
@ResidentMario
GitHub's links are all bluer now, it's really weird-looking to me: like I accidentally hit the saturation button on my monitor a bunch of times.
Paul Hobson
@phobson
@ResidentMario I use geopandas a bit and saw your geoplot module. very cool! I dig it.
Aleksey Bilogur
@ResidentMario
thank you for the kind words :smile:
Elliott Sales de Andrade
@QuLogic
heh, at first I didn't know what anyone was talking about with the links, but then I went to this page and the big heading+bright blue is super obvious now.
Paul Hobson
@phobson
0_o
Thomas A Caswell
@tacaswell
is appveyor always broken recently or am I just getting really unlucky seeing transients?
Paul Hobson
@phobson
I feel like it's mostly, but not always, broken
Elliott Sales de Andrade
@QuLogic
I see downloading freetype failed; can we put that in a cache somewhere?
Thomas A Caswell
@tacaswell
We definitely should cache that!
we do on travis, but I am not sure how caching works on appveyor though
Elliott Sales de Andrade
@QuLogic
I see there's no Downloading line in Travis, but I'm not sure where the caching is setup?
Oh, XDG cache dir; good choice then.
Adrien F. Vincent
@afvincent
About the new design of the GitHub homepage: https://www.designernews.co/stories/80826-the-new-github-homepage
Hetterick
@Hetterick
Hello there, I am very new here. For my thesis, I need to find a way to represent confusion matrices "nicely".
I would like to do something like this with scaling colors : https://i.stack.imgur.com/vOUcK.jpg
I could saw that this is not so difficult with matplotlib.
But then I would like to add the fact to have different areas of squares (for instance if have a categorie with much more data), something like this : https://inspirehep.net/record/1209003/files/CKM-PMNS.png
Do you think it is difficult to do? And if no, do you have a clue on how to do it?
Adrien F. Vincent
@afvincent
A quick thought: you might find some inspiration from the Hinton's diagram example in the gallery here.
Hetterick
@Hetterick
Thanks :)
I will take a look at it !
Adrien F. Vincent
@afvincent
eventplot_and_IndexLocator.png

Folks, I am not sure it can actually be considered as a bug but I find the behavior of IndexLocator quite surprising when it is coupled to a plot drawn with eventplot. For example, the following script

import numpy as np
import matplotlib.pyplot as plt
from matplotlib.ticker import IndexLocator

# Event sequence is "10" at 0.1 s, "21" at 0.5 s, and "10" at 0.9 s.
events = [[0.1, 0.9], [0.5]]
indices = [10, 21]
event_height = 0.9  # not 1 to better visualize rows of events

fig, axs = plt.subplots(ncols=2, num='eventplot_and_IndexLocator')
for ax in axs:
    ax.eventplot(events, lineoffsets=indices, linelengths=event_height)

# One wants the ticks to be all the indices: a possible thought may
# be to use an IndexLocator with base = 1 and offset = 0:
axs[0].yaxis.set_major_locator(IndexLocator(1, 0))
# Issue 1?: the tick values are not integers...

# Case that is even worse: let's assume that for any good reason, we want
# the ticks to start at 12 (offset = 12), and go 5 by 5 (base = 5).
axs[1].yaxis.set_major_locator(IndexLocator(5, 12))
# Issue 2?: one would expect the ticks to be 12, 17 and 21...

plt.tight_layout()  # for eyes' pleasure
plt.show()

produces the image above.

For the first case, MultipleLocator can be a working solution. However, it does not seem to provide the possibility to use an offset as IndexLocator does.
Antony Lee
@anntzer
@afvincent I would say it's a bug. In fact once it is fixed I can't really see why MultipleLocator and IndexLocator need to both exist, I'd suggest just adding an offset option to MultipleLocator and deprecate IndexLocator.
Adrien F. Vincent
@afvincent
I tried to fix IndexLocator for a while before realizing that MultipleLocator already implement (and in a better way ^^) most of what I was trying to achieve (through a Base class). So I would agree that “just” adding an offset kwarg to MultipleLocator seems a more appealing solution :). For the record, from what I understood, one of the reason to my IndexLocator problem is that this locator uses the “data limit” to work with (BTW, why rely on the data limits and not the view ones?), which does not return integer values in the case of an eventplot with linelengths != 1. Another part of the issue is that the the way the offset is handled does not take into account the case where offset >= base.
Benjamin Root
@WeatherGod
Eric Firing would probably be in the best position to answer your questions, but he is never on gitter. So, I suggest sending an email to the mailing list to get his attention.
Adrien F. Vincent
@afvincent
So many communication channels are available ^^. I was planning to open an issue on GitHub, but sending an email to the devs' mailing-list could indeed also be a solution.
David Stansby
@dstansby
Is there a way to restart just one Appveyor build instead of the whole job?
Elliott Sales de Andrade
@QuLogic
I haven't found one so far.
David Stansby
@dstansby
appveyor/ci#58
Adrien F. Vincent
@afvincent
They are missing our whishlist label :grin:!
Elliott Sales de Andrade
@QuLogic
Everyone should go :+1: that...
Nelle Varoquaux
@NelleV
I wonder why github decided that bright colors was better for usability…
hannah
@story645
Could also be branding
Nelle Varoquaux
@NelleV
it's definitely branding, but they also mention "Colors are higher contrast and more saturated to help make the site feel a little less old, improve readability for more people, and help address accessibility and colorblind issues" which makes me very confused considering bright colors (and too many colors) is known to decrease readability.
Or so I thought…
hannah
@story645
Well they're not talking about brightness... saturation is a different property
Nelle Varoquaux
@NelleV
eh
indeed my vocabulary is wrong
by that I mean it is not scientifically accurate.
hannah
@story645
shrugs I've just heard way too many lectures on color systems...
Nelle Varoquaux
@NelleV
I'll mention to my roomate that the painting she's trying to sell for 100k$ is to saturated to see the look on her face…
hannah
@story645
:)
Is it actually too saturated or too bright?
Nelle Varoquaux
@NelleV
probably both :(
hannah
@story645
Hmm that sounds like it could be cool in a pop art sort of way...
Adrien F. Vincent
@afvincent
@anntzer @WeatherGod Ok, I think I managed to properly add an offset kwarg to MultipleLocator (easier than what I was trying to do with IndexLocator… => ~ 3 slocs). I'll try to send an email (maybe only this weekend) about this on the mailing-list, hoping for Eric's attention. We'll see if it is worth a PR and maybe what to do with IndexLocator (, which appears to be a very old piece of the codebase).