Fix Segmentation fault on foreach_datapar_zipiter

[srinivasyadav18]

Proposed Changes

• The len variable in initialised to count but did not use rightly in the loop.
• replace count with len, must be a typo.

Any background context you want to provide?

#5235 (comment)

[jenkins-cscs]

Can one of the admins verify this patch?

[msimberg]

retest

[srinivasyadav18]

Here a2d7511, is_data_aligned is logically correct and is used in correct sense, but the convention was reverse, it returns false when data is aligned. I now changed it to return true when is aligned.

[msimberg]

retest

[hkaiser]

why do we the need of first value (false/true) in sequencer, when we are using only from 1 to sizeof...(Is) + 1? in std::all_of(), why can't we directly use from 0 to sizeof...(Is) without the first value in sequencer?,

We definitely could. Be careful to account for a zero length sequence, however.

[srinivasyadav18]

why do we the need of first value (false/true) in sequencer, when we are using only from 1 to sizeof...(Is) + 1? in std::all_of(), why can't we directly use from 0 to sizeof...(Is) without the first value in sequencer?,

We definitely could. Be careful to account for a zero length sequence, however.

@hkaiser Okay, then the existing solution is a better one. I did not realise about zero length sequence.
And could you please suggest anything about this #5281 (comment)
After adding or editing the tests, most probably there are no further issues related to datapar.

[srinivasyadav18]

@msimberg @hkaiser I have made the changes! Please mention if any further changes are required :-).

[msimberg]

@srinivasyadav18 I'll merge this into the other branch now, let's do further changes if needed there (I have nothing in mind right now).