Franck HOUSSEN
Franck HOUSSEN
`cmake` may end-up with different compile / link options than `configure`. AFAIR, when using intel, you need to make sure your env is intel-compliant : some scripts (`ifortvars.sh` or similar)...
You may propose a PR. If so, think about also providing / updating iso_c_ binding API.
BTW, not sure to get the pb right. Seeds are fixed: so generated random numbers should be the same if you provide the same initial vector, no ?
OK, I can try to PR something about that. So you need something to reset seed at every iteration of the RCI-while loop, right ?
> I have a branch here: master...pv:rng-seed Looks a bit fuzzy (many impacts) and incomplete (missing all p*getv0.f). Tried to PR some fix : not tested, @pv could you test...
Can't afford more time on this. Hope somebody can.
>There is indeed debug.h in the SRC file, but my pArpack installation has no headers. It's probably missing in cmake's scripts. Yes, I would say `debug.h` is missing, and, that...
Sure. I didn't tested that (you'll probably need to "play" with `extern` and `static` anyway...) : my first try would be to add " static debug_ dbg_ + initialize dbg_...
Oh men !... Just realized `debug.h` is fortran, not C (I was mislead by the file name). My suggestion will probably not work (at least not so "easily"...). I have...
I had a closer look at that. Seems not trivial at all !... For now, I have no solution. I'll PR one if I get something to work.