[Rd] re sponses to issues on r-devel

Ben Bolker bolker at ufl.edu
Fri Nov 16 16:02:30 CET 2007



  I'm probably going to get in trouble for this, but ... 

  What is the best way to get 'important' issues resolved
by R-core?  There is a category of 'issues' (I won't call
them bugs or problems, although arguably they are so)
that appear on R-devel and sink without ever receiving an
explicit acknowledgment.  These problems tend to be:
 * non-trivial 
 * outside of the main current interests (perhaps) of R-core

  I'm thinking in this case of Petr Savicky's series of posts

http://www.nabble.com/bug-in-detection-of-zero-state-for-Mersenne-Twister-%28PR-10362%29-tf4670916.html#a13343487
http://www.nabble.com/initial-scrambling-of-seed-in-do_setseed---RNG_Init-tf4504008.html#a12845234
http://www.nabble.com/all-zeroes-in-Mersenne-Twister-state-may-remain-undetected-tf4626354.html#a13210502
http://www.nabble.com/predictable-bit-patterns-in-runif%28n%29-shortly-after-set.seed-tf4641681.html#a13257396

all between Sept 23 and October 15.   One of these was posted as a bug, but
the rest are just
"ideas".  However, they are very thoroughly documented ideas (with suggested
patches),
and it would be a shame if they fell between the cracks.  They might be on
R-core's "to do" list already  -- or
someone may have replied off-list -- or R-core might have read the posts and
decided
that they are not worth worrying about -- or R-core might go through the
archives
looking for these kinds of unresolved issues from time to time.

   Arguably the best solution to this kind of problem is just persistence by
the 'R-cortex' (OED:
 {dag}1. fig. The external part; the outer shell or husk. Obs.) [by which I
mean the people like
Petr who are capable of finding, diagnosing and creating useful patches],
re-posting
to R-devel every month or so if issues remain unresolved.  My *personal*
feeling
is that posting these kinds of issues as bugs, where they can be remembered
and dealt with,
is a good idea, but I'm prepared to be shouted down ...

  Is this a problem?  Is there a solution?

   (Please don't blame Petr for these comments, I'm just using his ideas as
an example)

   cheers
    Ben Bolker
-- 
View this message in context: http://www.nabble.com/responses-to-issues-on-r-devel-tf4806780.html#a13752138
Sent from the R devel mailing list archive at Nabble.com.



More information about the R-devel mailing list