From maechler at stat.math.ethz.ch  Mon Mar 10 10:43:20 2025
From: maechler at stat.math.ethz.ch (Martin Maechler)
Date: Mon, 10 Mar 2025 10:43:20 +0100
Subject: [Expm-developers] Follow-up
In-Reply-To: <add06a10-708b-493e-a41f-f015da28d334@mayo.edu>
References: <add06a10-708b-493e-a41f-f015da28d334@mayo.edu>
Message-ID: <26574.46136.314163.812785@stat.math.ethz.ch>

>>>>> Therneau, Terry M , Ph D 
>>>>>     on Sat, 08 Mar 2025 14:19:56 -0600 writes:

    > By trial and error experimentation I have the remote link to expm working.? Change void * 
    > to DL_FUNC in init.c, and? to extern DL_FUNC in upper.c, which is the the function makes 
    > use of expm and matexp_MH09 calls.??? Interestingly, the? ubuntu box without a Makevars 
    > gives a compiler warning, but the one with

    > ?? CFLAGS=? -g -O2 -Wall -pedantic -mtune=native -fno-common -Wstrict-prototypes

    > is happy.?? I will have to send another email with that warning message, since I can't log 
    > into that one remotely.?? The installed library runs on both.

    > Summary:
    > ?I seem to have solved the immediate issue.? This is now purely FYI information.? I am 
    > still puzzed by the different behaviors, but it is only curiousity.

    > Terry T.

    > Martin: I cc'd this to you, only in case the expm-developers email (where I sent the prior 
    > note) is stale.


    > init.c:
    > #include "R.h"
    > #include "R_ext/Rdynload.h"

    > /* Interface to expm package. */
    > typedef enum {Ward_2, Ward_1, Ward_buggy_octave} precond_type;
    > DL_FUNC (*expm)(double *x, int n, double *z, precond_type precond_kind);
    > DL_FUNC (*matexp_MH09)(double *x, int n, const int p, double *ret);
    > void R_init_hmm(DllInfo *dll)
    > {
    > ??? expm = (DL_FUNC) R_GetCCallable("expm", "expm");
    > ??? matexp_MH09= (DL_FUNC) R_GetCCallable("matexp_MH09", "matexp_MH09");
    > }


Dear Terry,
thank you for this.

When* did you send a previous e-mail?
I may have overlooked (I'm embarrassed to say that this happens
regularly to me; the main reason that I'm getting too many
e-mails ( <== am involved into too many projects).

Are suggesting a change to the expm package sources such that
whatever problem you had  would _not_ have been necessary to be
solved on your side ("downstream") ?

Best regards,
Martin

    > -- 
    > Terry M Therneau, PhD
    > Department of Quantitative Health Sciences
    > Mayo Clinic
    > therneau at mayo.edu

    > "TERR-ree THUR-noh"

    > [[alternative HTML version deleted]]