Ryan Murray

Wow, this is urgent. I suppose these have always been unofficial binaries that people take for granted. Porting the wheels isn't a long term solution since it'll freeze the builds...

The url fetcher sorts alphabetically, which is why 0.6 comes before 0.10. See here: https://github.com/lepisma/pipwin/blob/acd86ad128e9624502ec35e686df3184abbc49aa/pipwin/pipwin.py#L260

Hi @lepisma ... I'd be happy to but I don't have time anytime soon. So please feel free anyone else in the meantime

I don't know how exactly how it's used in every DFS file, but it seems it's at least possible: http://docs.mikepoweredbydhi.com/core_libraries/dfs/dfs-file-system/#custom-blocks I'm mostly familiar with DFS2 custom blocks. For example, the...

Haha, yea it's a long time to match a big file, but hopefully it could be improved nonetheless :) . > * Number of timesteps _1 timestep (it's a statistics...

Here's the profile for testing it on a much smaller dfsu (report attached as zip). I'll try a test on a larger file overnight to see how scale might change...

The results from the larger dfsu file are in (no. elements = 3 760 848). It looks like the contains() method is the main culprit. The block taking ~15% is...

> Did you try > > ``` > da_grid = da.interp_like(g, extrapolation=True) > ``` This definitely helped a lot, thanks! The final result still needs masking though, I'll try it...

Nice! I think that's sufficient for writing blocks. Is it also possible to read the values?

Similar relevant use case: If we want to import cross section data which comes in pairs of (distance from the left, elevation), first we have to create a cross section...