Dirk, That was indeed off-the-cuff noise and not intended as a report. I hadn't had cause to build Rcpp since mid-November, at which time I did receive a timeout (I could be recalling incorrectly on the step though) so I'm glad to hear that the test time has been reduced and pleased to see the results.
Thanks for the heads-up on ccache, it'll be interesting to see how it plays with clang. On Sun, Dec 16, 2012 at 11:53 AM, Dirk Eddelbuettel <e...@debian.org> wrote: > > On 16 December 2012 at 11:23, Thell Fowler wrote: > | endings in src/Timer.cpp , and jss.bst already being included in R. I'd > | mention further but doRunit.R usually times out on my system and I don't > > Huh? You mean you run out of patience? It always finishes. > > There were times when due to coding errors the headers recursed too much, I > have seen this to be on the order 10+ minutes. But that was a few weeks > ago. > > Right now it is pretty reliably around 200 secs for the unit tests. As > previously shown, I prefix gcc and g++ with ccache [ which is close to > being > the coolest thing since sliced bread ] so I may gain a little there. The > timings are relative to the tarball from R CMD build and with timings > enables > in ~/.R/check.Renviron. > > So if you have something to report, please make it reproducible (to the > extent that compilers etc are affected -- I don;t have clang3.3 yet) rather > than an off-the-cuff remark which is just helpful. Patches, as always, are > the true path to glory and you did that :) > > Dirk > > -- > Dirk Eddelbuettel | e...@debian.org | http://dirk.eddelbuettel.com > -- Sincerely, Thell
_______________________________________________ Rcpp-devel mailing list Rcpp-devel@lists.r-forge.r-project.org https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
