I recently changed the back-end compiled code for the Delaporte
package from C++ to Fortran/C, and after some birthing pains (and a
lot of patience shown by CRAN), it's fully functional _except_ on
sparc-sun-solaris2.10 [1]. There, 10 of the 44 tests return errors.
I've done enough research to know that there may be endian issues
involved, but I do not know how to address them.
Interestingly, some of the errors look as if they are caused by
returning the "ones-complement" (if that is a word, meaning returning
1 - x instead of x) in some cases. See tests 3, 4, 6, and 7 for
examples. In the next version (which won't be posted to CRAN for at
least a month 8-) ), I've expanded the number of tests from 44 to (at
least) 86 to try to more cleanly isolate possible failures, but I
don't have access to a sparc-based Sun server, nor is there one on
Rhub and win-builder, so I cannot test it.
Has anyone seen anything similar or does this ring any bells?
Thank you,
Avi
[1] <
https://www.r-project.org/nosvn/R.check/r-patched-solaris-sparc/Delaporte-00check.html