Create seq class for C++ API

[shehzan10]

Implement seq class for C++ API. Move C-API seq to af_seq.
Implement span and end.
Refactor code from v2.1.

[hakttari]

Hello,
The following code is working with AF 2.0, but it makes an error in AF 3.0. How can I implement af::seq with arrayfire open source library?

int m = 5;
array i = seq(m);

[9prady9]

@hakttari I think we missed this functionality during the code refactor. I have created an feature request for this. You can follow up on progress here

[shehzan10]

This functionality is replaced by iota. The function definition are shown below.
The rep argument allows a repetition dimension to be chosen.

AFAPI array iota(const dim4 &dims, const unsigned rep = 3, af_dtype ty=f32);
    AFAPI array iota(const dim_type d0, const unsigned rep = 0, af_dtype ty=f32);
    AFAPI array iota(const dim_type d0, const dim_type d1,
                     const unsigned rep = 1, af_dtype ty=f32);
    AFAPI array iota(const dim_type d0, const dim_type d1, const dim_type d2,
                     const unsigned rep = 2, af_dtype ty=f32);
    AFAPI array iota(const dim_type d0, const dim_type d1, const dim_type d2,
                     const dim_type d3, const unsigned rep = 3, af_dtype ty=f32);

[umar456]

What's the point of taking it out? Why shouldn't we put it back in?

Sent from my iPhone

On Dec 18, 2014, at 11:52 PM, Shehzan Mohammed notifications@github.com
wrote:

This functionality is replaced by iota. The function definition are shown
below.
The rep argument allows a repetition dimension to be chosen.

AFAPI array iota(const dim4 &dims, const unsigned rep = 3, af_dtype ty=f32);
AFAPI array iota(const dim_type d0, const unsigned rep = 0,
af_dtype ty=f32);
AFAPI array iota(const dim_type d0, const dim_type d1,
const unsigned rep = 1, af_dtype ty=f32);
AFAPI array iota(const dim_type d0, const dim_type d1, const dim_type d2,
const unsigned rep = 2, af_dtype ty=f32);
AFAPI array iota(const dim_type d0, const dim_type d1, const dim_type d2,
const dim_type d3, const unsigned rep = 3,
af_dtype ty=f32);

—
Reply to this email directly or view it on GitHub
#51 (comment).

[pavanky]

iota is more complete than seq. We can have a wrapper for this particular operation, but ideally I'd prefer if we forced the usage iota as opposed to seq.

[umar456]

We shouldn't break things unless there is a good reason.

Sent from my iPhone

On Dec 18, 2014, at 11:57 PM, Pavan Yalamanchili notifications@github.com
wrote:

iota is more complete than seq. We can have a wrapper for this particular
operation, but ideally I'd prefer if we forced the usage iota as opposed to
seq.

—
Reply to this email directly or view it on GitHub
#51 (comment).

[pavanky]

@shehzan10 Can you create a wrapper for this to use iota internally ?

[pavanky]

@hakttari This has been added to the current devel branch.