Sean Gillies

I deleted some irrelevant comments. Please don't do this, people. Use the discussion group linked in the README under "support".

@thomascoquet I'll try the code soon. Meanwhile, what do you see if you replace `array = np.ones((5, 5))` with `array = np.ones((5, 5), dtype="uint8")`? I confess I know very little...

@thomascoquet thanks for the reply. I've tagged this as a GDAL issue upstream. I feel like GDAL should raise an exception if it can't accept the requested parameters instead of...

@wckoeppen no movement. I would have closed this issue if it were resolved or you would see other issues pointing here if there was any discussion.

@skogler please don't. I haven't changed my position from #2200. Rasterio is very different from urllib3. It's nowhere near as mature. And Python typing itself is not fully mature.

A little more explanation. Rasterio's biggest problem is not the lack of type hints. The biggest problem is complexity. We have to deal with the ever-increasing complexity of GDAL and...

As an example of how relatively immature rasterio is, we (and by we I mean me) haven't done a great job at documenting rasterio's design. I fleshed out https://github.com/rasterio/rasterio/blob/master/CONTRIBUTING.rst#design-principles and...

@lytk01 I'm inclined not to surface GDAL's statistics API in Rasterio because GDAL's statistics are limited in comparison to Numpy's: https://docs.scipy.org/doc/numpy-1.15.1/reference/routines.statistics.html. Instead, I'd like to offer examples or perhaps a...

I'm going to leave this issue open as a reminder to start the discussion. We'll close it after.

@jhamman GDAL requires a `GByte *` (`unsigned char *`) buffer: https://gdal.org/api/cpl.html#_CPPv420VSIFileFromMemBufferPKcP5GByte12vsi_l_offseti, hence we read file objects entirely into memory and then provide that buffer to GDAL.