I haven't played with ndslice nor followed its deprecation discussions. Could someone summarize it for us please. Also, is it still used outside Phobos or is Ilya or someone else rewriting it?

Ali

On Thursday, 13 April 2017 at 08:47:16 UTC, Ali Çehreli wrote:
> I haven't played with ndslice nor followed its deprecation discussions. Could someone summarize it for us please. Also, is it still used outside Phobos or is Ilya or someone else rewriting it?
>
> Ali

Hello Ali,

ndslice was removed from Phobos because it is to hard to maintain Phobos and Mir at the same time. It is better to have dub packages instead of big Phobos, IMHO.

ndslice was completely rewritten and extended in mir-algorithm package [0, 1]  since ndslice was deprecated in Phobos. Its API is stable enough and includes all kinds of tensors. New ndslice is used in two Tamediadigital's projects [3, 4]. See also its README for more details.

The old ndslice (like in Phobos) is located in parent Mir package [2].
The last Mir version with old ndslice is v0.22.1.

[0] http://docs.algorithm.dlang.io
[1] https://github.com/libmir/mir-algorithm
[2] https://github.com/libmir/mir
[3] https://github.com/tamediadigital/lincount
[4] https://github.com/tamediadigital/hll-d

If you have any questions about ndslice I would be happy to answer.

Best regards,
Ilya

On Thursday, 13 April 2017 at 08:47:16 UTC, Ali Çehreli wrote:
> I haven't played with ndslice nor followed its deprecation discussions. Could someone summarize it for us please. Also, is it still used outside Phobos or is Ilya or someone else rewriting it?
>
> Ali

The reasons to use mir-algorithm instead of std.range, std.algorithm, std.functional (when applicable):

1. It allows easily construct one and multidimensional random access ranges. You may compare `bitwise` implementation in mir-algorithm and Phobos. Mir's version few times smaller and do not have Phobos bugs like non mutable `front`. See also `bitpack`.
2. Mir devs are very cary about BetterC
3. Slice is universal, full featured, and multidimensional random access range. All RARs can be expressed through generic Slice struct.
4. It is faster to compile and generates less templates bloat.
For example:

slice.map!fun1.map!fun2

is the same as

slice.map!(pipe!(fun1, fun2))

`map` and `pipe` are from mir-algorithm.

On Thursday, 13 April 2017 at 10:00:43 UTC, 9il wrote:
> On Thursday, 13 April 2017 at 08:47:16 UTC, Ali Çehreli wrote:
>> I haven't played with ndslice nor followed its deprecation discussions. Could someone summarize it for us please. Also, is it still used outside Phobos or is Ilya or someone else rewriting it?
>>
>> Ali
>
> The reasons to use mir-algorithm instead of std.range, std.algorithm, std.functional (when applicable):
>
> 1. It allows easily construct one and multidimensional random access ranges. You may compare `bitwise` implementation in mir-algorithm and Phobos. Mir's version few times smaller and do not have Phobos bugs like non mutable `front`. See also `bitpack`.
> 2. Mir devs are very cary about BetterC
> 3. Slice is universal, full featured, and multidimensional random access range. All RARs can be expressed through generic Slice struct.
> 4. It is faster to compile and generates less templates bloat.
> For example:
>
> slice.map!fun1.map!fun2
>
> is the same as
>
> slice.map!(pipe!(fun1, fun2))
>
> `map` and `pipe` are from mir-algorithm.

It is all good, but I am sure many D programmers, myself included, would appreciate if shortcomings of Phobos are fixed instead of having a completely separate package with set of features that overlap... I understand ndslice was at some point in the `experimental` package, but again - it would be good if you improve existing Phobos stuff instead of providing a separate library that provides better implementation(s).

On Thursday, 13 April 2017 at 15:00:16 UTC, Dejan Lekic wrote:
> On Thursday, 13 April 2017 at 10:00:43 UTC, 9il wrote:
>> On Thursday, 13 April 2017 at 08:47:16 UTC, Ali Çehreli wrote:
>>> [...]
>>
>> The reasons to use mir-algorithm instead of std.range, std.algorithm, std.functional (when applicable):
>>
>> 1. It allows easily construct one and multidimensional random access ranges. You may compare `bitwise` implementation in mir-algorithm and Phobos. Mir's version few times smaller and do not have Phobos bugs like non mutable `front`. See also `bitpack`.
>> 2. Mir devs are very cary about BetterC
>> 3. Slice is universal, full featured, and multidimensional random access range. All RARs can be expressed through generic Slice struct.
>> 4. It is faster to compile and generates less templates bloat.
>> For example:
>>
>> slice.map!fun1.map!fun2
>>
>> is the same as
>>
>> slice.map!(pipe!(fun1, fun2))
>>
>> `map` and `pipe` are from mir-algorithm.
>
> It is all good, but I am sure many D programmers, myself included, would appreciate if shortcomings of Phobos are fixed instead of having a completely separate package with set of features that overlap... I understand ndslice was at some point in the `experimental` package, but again - it would be good if you improve existing Phobos stuff instead of providing a separate library that provides better implementation(s).

Work on Phobos is useless for me because when I need something and I am ready to write / fix it I have few days, but not few months until LDC release. DUB is more flexible, it allows to override version with local path for example. Finally, I think Phobos should be deprecated and be split into dub packages.

On Thursday, 13 April 2017 at 08:47:16 UTC, Ali Çehreli wrote:
> I haven't played with ndslice nor followed its deprecation discussions. Could someone summarize it for us please. Also, is it still used outside Phobos or is Ilya or someone else rewriting it?
>
> Ali

We should additionally mention sometimes, that the naming of ndslice is derived from ndarray
and this is from N-dimensional Array. No one searching for N - dimensional Array OR Matrix will find ndslice, without this info easily. Just try a google search: "dlang n dimesional array"
Regards mt.

On Thursday, 13 April 2017 at 15:22:46 UTC, Martin Tschierschke wrote:
> On Thursday, 13 April 2017 at 08:47:16 UTC, Ali Çehreli wrote:
>> I haven't played with ndslice nor followed its deprecation discussions. Could someone summarize it for us please. Also, is it still used outside Phobos or is Ilya or someone else rewriting it?
>>
>> Ali
>
> We should additionally mention sometimes, that the naming of ndslice is derived from ndarray
> and this is from N-dimensional Array. No one searching for N - dimensional Array OR Matrix will find ndslice, without this info easily. Just try a google search: "dlang n dimesional array"
> Regards mt.

The first link for me in incognito mode is:
https://wiki.dlang.org/Dense_multidimensional_arrays

It contained example for std.experimental.ndslice. But thanks for the idea. Reworked  for Mir now.

Thanks,
Ilya

On Thursday, 13 April 2017 at 15:22:46 UTC, Martin Tschierschke wrote:
> On Thursday, 13 April 2017 at 08:47:16 UTC, Ali Çehreli wrote:
>> I haven't played with ndslice nor followed its deprecation discussions. Could someone summarize it for us please. Also, is it still used outside Phobos or is Ilya or someone else rewriting it?
>>
>> Ali
>
> We should additionally mention sometimes, that the naming of ndslice is derived from ndarray
> and this is from N-dimensional Array. No one searching for N - dimensional Array OR Matrix will find ndslice, without this info easily. Just try a google search: "dlang n dimesional array"
> Regards mt.

... plus link in the spec
https://github.com/dlang/dlang.org/pull/1634