New syntax for Cube extraction

[koldunovn]

I just recently begin to use Iris, and enjoy it very much. It has a lot of promising features. However your syntax for "Cube extraction" just kill all the fun. Cutting different regions and periods from the data is very basic operation during data analysis, and doing this with a lot of lambda functions is neither fast, nor convenient in practice. If for cutting simple box from my data I have to do something like this:

latc = iris.Constraint(latitude=lambda cell: minlat <= cell <= maxlat)
lonc = iris.Constraint(longitude=lambda cell: minlon <= cell <= maxlon)
region = cube.extract(latc & lonc)

it makes me very very sad.

I just wounder if you might be interested in discussion about how to do this in a better way. I see that there were some suggestions made in Issue #77 , but I think this topic deserves it's own ticket :)

Let me start with my humble suggestions, so that you have something to criticize :) I would love to have syntax similar to Pandas, where you can explicitly use values of your dimensions to limit the data. Something like:

cube.ext( t=['1990-01-01':'2000-01-01'], x=['140E':'160W'], y=['10S':'20N'], z=['0':'2000'] )

This, as you might notice, resembles also the Ferret syntax, that I believe is very convenient, and would be useful to have. I understand that the cube can have more dimensions, but having this four covered would be already a great deal. I do not suggest to replace existent syntax with this one, but only to create additional method for the cube class (e.g. ext), that will allow this type of syntax for extraction of the region.

What do you think?

[rcomer]

Hi @koldunovn, I’m just reviewing some old issues to see what still needs addressing. It looks like the cube.intersection method fits your suggestion for latitude and longitude at least. Time extraction has also moved on and become less painful since you opened this. So I’m going to close this now. If you are still using Iris and think there is more to do here then please feel free to open a new issue.