Re: [Pytables-users] `__iter__` state and `itertools.islice` when

[Antonio V. <ant...@ti...>]

Hi Tony,


Il giorno 09/lug/2013, alle ore 06:38, Tony Yu <ts...@gm...> ha scritto:

> Hi,
> 
> I ran into a subtle, unexpected issue while using `itertools.islice`. I wanted to pass slices of an array for processing without actually reading the entire array, and I wanted that processing function to know nothing about how I'm taking that slice. To that end, I had a loop that sliced the array using `itertools.islice` and called the function on each slice. Instead of returning the slice I specified, `islice` treated the previous end slice as the starting point to the next slice.
> 
> That description is a bit confusing, but the example below (along with the attached test data) should illustrate the point. Maybe I'm missing something, but the only work around that I found was to set a private flag (e.g. `h5.root.array._init = False`) on each call to `islice` to reset the counter used in `__iter__`.
> 
> I'm not sure if this is expected behavior or not, but it does differ from how `islice` works on numpy arrays (as demonstrated in the example below). I used the google and nothing similar came up, so I thought I'd post here.
> 
> Best,
> -Tony 
> 
> 
> #~~~~
> 
> import tables
> import itertools
> import numpy as np
> 
> 
> h5 = tables.openFile('test.h5')
> array = np.arange(100)
> for i in range(5):
>     # Numpy array slice always returns 0..10
>     print list(itertools.islice(array, 0, 10))
>     # PyTables array slice shifts with each iteration
>     print list(itertools.islice(h5.root.array, 0, 10))
> h5.close()
> <test.h5>------------------------------------------------------------------------------


Yes, this is a bug IMO.
Thank you for reporting and thank you for the small demonstration script.

Can you please file a bug report on github [1]?
Please also add info about the PyTables version you used for the test..



[1] https://github.com/PyTables/PyTables/issues

--
Antonio Valentino