Stan in R on PowerPC G5 (10.6 PPC): Help needed

UPD. Deleted lgamma issue, since got a reply, and also found a relevant passage in documentation. Likely no longer relevant (cannot test right now, implementation of tests in portfile is incomplete so far – I have been busy fixing OCaml build system).

So far OCaml looks promising – packages mostly build smoothly, though build system is alien and hard to tweak. Will update once I reach Stanc. There are many dependencies not just to build, but to implement in portfiles: Macports is lacking some 50+ relevant packages.

I am afraid that ocamlc gonna be slow though, and on G4 horribly so. (Well, maybe with Stanc parallelization gonna save the case somewhat; at the moment ocaml-num compiles, and it reminded me of Guile, which takes some 15+ hours to boostrap tree dozen small files LOL.)

A quick update on current state of things (I may repeat myself, but I think it is nevertheless meaningful atm):

I have basically fixed oneTBB and all Rcpp* dependencies. Since Macports has no easy API for installing R packages – and I need to fix broken ones – I have written my own.

stanc3 builds with no issues at all, and I have just built rstan. Need to rebuild it again, since it suffers from the same malloc errors like all Rcpp* ports. But I know how to fix that. Just gonna take another hour+, since I am away from the Quad, and my other PowerMac is temporarily with a very little RAM, so everything takes time.

On a positive side, maybe we gonna have the full R environment in Macports soon.

@andrjohns

Seems like working! (May need to set DYLD_LIBRARY_PATH for model compilations too to fix malloc errors.)

1 Like