12:31 < mef> xen48# nm  /usr/lib/libm.so |grep log1pl
12:31 < mef> xen48# nm  /usr/lib/libm.so |grep log1p 
12:31 < mef> 000000000001def0 T log1p
12:31 < mef> 0000000000018790 T log1pf
12:31 < mef> I'm looking for the symbol (function) log1pl
12:32 < mef> where is it ?
12:34 < mef>  grep -r log1pl  /usr/include/ /usr/pkg/include
12:34 < mef> may answer to my question
12:36 < mef> in /usr/include/math.h, there is a line,  but not in /usr/lib/libm.so (?)
12:50 < mef> #ifdef notyet // missing log1pl __HAVE_LONG_DOUBLE
12:51 < mef> math/R-igraph (1.3.1) wants this log1pl ;-(
12:54 < mef> R-nimble-0.12.2 wants R-igraph (1.3.1)

00:51 < tnn> mef: I think you can just do #define log1pl(x) logl(1.0f+(x)) under #ifdef __NetBSD__ and ignore ieee epsilon semantics
00:52 < tnn> there are other parts of libm in NetBSD that are not IEEE compliant + no comprehensive regression test suite. Anyone who seriously relies on it for scientific use will be disappointed.

06:44 < mef> xen48@makoto 06:43:56/220427(..textproc/R-rmarkdown)% emacs /tmp/1                                            
06:44 < mef> emacs: Undefined PLT symbol "__sigaction_siginfo" (symnum = 331
06:45 < mef> (emacs26, I've compiled on 9.99.93, but running on 9.99.82 ;-(, I'll update the latter)
06:45 < mef> Is that correct observation ?

(cvs update as of 2022/04/10) on 9.99.93 host:

h97@makoto 13:15:28/220426(..lib/libm)% nm /usr/lib/libc.so |grep sigaction_sigi
000000000019c052 T __sigaction_siginfo