Export C level forder for use by other packages' C code

[mattfidler]

Hi,

I was wondering if you would consider exporting forder for use in other packages?

I can create a pull request if you think this is reasonable.

Best,

Matt.

[jangorecki]

You are welcome to submit PR. I will close this issue as it was previously reported in #3447. Please read linked discussion there.

[mattfidler]

That is not exactly what I meant. I mean C-level export. It should be clear in the pull request.

[jangorecki]

If you ask for C interface, then agree it is not a duplicate.

[mattfidler]

Yes. Rereading my original issue I found it is ambiguous.

My purpose was to use radix sorting internally on a single thread; Reading the code closely I'm unsure this is easily possible and would be an edge case.

If you wish to ignore or close this issue, it would be OK with me.

[MichaelChirico]

isn't single threaded radix sorting just the same as what base R is doing? the advantage of including the data.table routine was precisely the parallelization, I thought?

…

On Fri, Nov 1, 2019, 4:31 AM Matthew Fidler ***@***.***> wrote: Yes. Rereading my original issue I found it is ambiguous. My purpose was to use radix sorting internally on a single thread; Reading the code closely I'm unsure this is easily possible and would be an edge case. If you wish to ignore or close this issue, it would be OK with me. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#4015?email_source=notifications&email_token=AB2BA5MHHWZC7QXVS6Q43NLQRM6DJA5CNFSM4JHKMDD2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOECZE5CA#issuecomment-548556424>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB2BA5NFJHQYHWH6QJQMWHDQRM6DJANCNFSM4JHKMDDQ> .

[mattfidler]

Unfortunately the C routines are not exported.

Indeed the advantage is parallelization. However I'm also unsure if their routines are thread safe (I assume they are) and I would like to sort on a single thread in a multi-threaded ODE solving where some of the solving times change based on model conditions.

The code isn't well commented, so I would have to assume authorship or say "R Core Team" if I simply duplicate what they have.

[mattfidler]

I do believe this was contributed by the data.table team too, so I'm unsure who to include from there.