This looks great! Thanks for the addition to the gallery.

Is there a way in the examples to more clearly show how multiple locator differes from maxN differed from auto? Like maybe change to a 1-10 range to have more wiggle room? Or use different scales-dunno, but I figure this image is at its most useful if readers don't necessarily have to read the code to get what it's trying to show.

@story645 You're right. Maybe 5 was not the best value to illustrate the different strategies. I'll try to play a bit more with the MaxN.

Yay! Now maybe something with multiples of 2 for multiple locator? And possibly using the axis text to explain what the formatters are labeling as that might explain strformatter vs scaler formatter. Formatstrformatter and strmethod formatter seem to work exactly the same/do the same thing...so I'm confused there.

FormatStrFormatter expects the old-style '%' based strings, StrMethodFormatter takes the new-style '{}' based format strings.

Ugh...since python documentations says both of those are format strings, wouldn't it have made more sense to fold the functionality for both into one of those functions? For the same of this example though, then maybe add the format string to the label FormatStrFormatter('>%d<')?

@rougier IndexLocator seems to be MultipleLocator where the start value for the ticks isn't necessarily 0. So given plot(range(2, 20)) Multiplelocator with base=5 will yield 5, 10, 15, 20 whereas indexLocator can be given base=5 and offset=2 and therefore yield 2, 7, 12, 17. This is probably most useful for year plots where you want every 5 or 10.

This might also work with just base=5, as the main point is IndexLocator locates every 5th point (hence index) whereas MultipleLocator locates multiples of 5.

@story645 And you need some data (i.e. some plot) to make it work, right ?

Not exactly, I think that's where using the offset argument may come in.

@story645 Coudl make a PR on my PR to show the IndexLocator case ?

Instead of a PR, I can give you code:

ax.plot(range(5, 55), [0]*50, color='White')
ax.xaxis.set_major_locator(ticker.IndexLocator(base=5, offset=2))
ax.text(0.0, 0.25, "IndexLocator (base=5, offset=2)", fontsize=16, transform=ax.transAxes)

Since IndexLocator is counting off ticks, it only works when data gets plotted.

Thanks, I've updated the PR and the screenshot at the top.

I still think you should change some of the other examples-like change multiple locator to be multiples of 5 or something so it's really clear what it does. And possibly put the parameters in the text for each of the locators/formatters.

Done.

Awesome! Can you also please add the name of the args for LinearLocator and LogLocator?
(And I fully support you changing the IndexLocator code to something like base=0.5, offset=.25 if you want to keep everything on more or less the same scale)

Also, can you put the format strings you used for FormatStr formatter and for StrMethod formatter in parenthesis next to each, so FormatStr formatter(">%d<") and the like?

Yay! A few slight nitpicks (I'm almost sorry for this, but...):

1. please space out the formatters a bit since it's getting crowded (yes, my fault on account of wanting the args)
2. add the arg name to LinearLocator
3. reorder the Locators and Formatters to either match the docs or so that the ones that don't take args come first (so Auto, Null, then in increasing order of complexity and with similar ones grouped together)

No problem but at this point I think it's a matter of personal test:

1. I think it's not too crowded
2. There is no arg name in my call to LinearLocator
3. It's only an example, so unless we want it somewhere in the doc, I think it's fine now, don't you think ?

1. Compromise? For FixedLocator and FixedFormatter, can you space out the text so it follows PEP8 coding style about spaces after commas?
2. numticks is the name of the first arg, and I think writing numticks=3 more explicitly gets across what's going on
3. shrugs personal preference, but I think ordering goes a long way towards clarity

1. Arrrggghhhh.... Ok, I'll do it even if I'm PEP8 allergic (cc @NelleV)
2. Makes sense
3. But more work for me (=PR appreciated)

Done. Didn't update screenshots at top.

Yay! Mind updating the picture?

I don't understand why FixedFormatter starts with a blank string, but the first tick still has a 0 on it?

+1 on @QuLogic's suggestion of using plt.subplots 'cause then you can just loop over the setup ax stuff trivially and that leaves the examples a lot cleaner and it becomes much more trivial to reorder them ;)

I think it's preferable to have subplot because it makes each sub-example self-contained. You can just copy and paste into your code and for me it's what example are useful for. The main point here are tick locators / formatters, it is not about subplot/subplots.

As for the empty string in FixedFormatter, I think this is a bug but I forgot to report.

I opened issue #7122 about the FixedFormatter problem. In the meantime, it seems there's also a bug in MultipleLocator (same issue). PR #7123 has been proposed to fix the bug.

Are we waiting for #7122 and #7123 to be fixed or can we merge as is?

Merge this as-is, clean up the fall-out later if needed.

Thanks @rougier !

backported to v2.x as 45d54e9