[tlambert03]

thanks @cvanelteren (and @beskep for the context in Ultraplot/UltraPlot#491).

I'd think I'd rather not register the entire cmap catlog with mpl on import: cmap can end up imported as a dependency of some other library, so an mpl user could have it in their session without knowing, and suddenly see ~1300 extra entries in plt.colormaps(), the Qt figure options dropdown, etc. Not to mention it would increase cmap import time 100 fold (from ~12 ms to 1 sec). (By contrast, cmasher only registers ~100)

So instead, this adds an explicit function:

import cmap

cmap.register_mpl("tol:*", "batlow")  # names or fnmatch patterns

plt.imshow(data, cmap="tol:bright")
plt.imshow(data, cmap="crameri:batlow_r")  # _r versions too

It only registers fully-qualified namespace:name keys and skips anything already registered. Names are canonical (lowercase, _ not -), so tol:bright_alt, not tol:bright-alt. I checked it with ultraplot 2.7.1: after cmap.register_mpl("tol:bright") or cmap.register_mpl("tol:*"), uplt.rc["cmap.qualitative"] = "tol:bright" works (without it, it raises a KeyError).

I'd caution against running cmap.register_mpl("*") at your own import time. If you'd rather not require users to call anything, another option on the UltraPlot side would be a lazy fallback when a key isn't found... something like this (untested sketch) in ColormapDatabase.__getitem__:

if not self._has_item(key):
    try:
        import cmap

        self.register(cmap.Colormap(key).to_mpl(), name=key)
    except (ImportError, ValueError):
        pass  # fall through to the usual KeyError

Thoughts? Does this solve your needs? Or are you really hoping for the full cmap catalog to be registered?

[codecov]

Codecov Report

❌ Patch coverage is 95.45455% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 95.90%. Comparing base (a9eab0b) to head (9fabe14).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
src/cmap/_util.py	95.23%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #162      +/-   ##
==========================================
- Coverage   95.91%   95.90%   -0.01%     
==========================================
  Files         168      168              
  Lines        2252     2273      +21     
==========================================
+ Hits         2160     2180      +20     
- Misses         92       93       +1     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[cvanelteren]

Hi @tlambert03 thanks for the quick turn around. I think for now this would suffice. We at ultraplot can register colormaps lazily, but matplotlib does not have this feature. I was not aware that the registration was so big on cmap's end! One way around the registration taking long would be to allow a lazy subclass to exist

from matplotlib.colors import Colormap

class LazyColormap(Colormap):
    def __init__(self, name, factory):
        super().__init__(name)
        self._factory = factory
        self._real = None

    def _load(self):
        if self._real is None:
            self._real = self._factory()
        return self._real

    def __call__(self, X, alpha=None, bytes=False):
        return self._load()(X, alpha=alpha, bytes=bytes)

Then this proxy would resolve on load. I think this may mittigate the loading taking long on cmap's end and should be compatible with mpl and ultraplot.