[Bioc-devel] Some info for maintainers

[Gordon Smyth]

Hi Jeff,

At 01:12 AM 21/10/2004, Jeff Gentry wrote:
>Hi there ...
>
>First I'd like to remind folks to keep themselves intimately familiar with
>the package checking results available
>at: http://www.bioconductor.org/checkResults/checkResults.html as that is
>what is happening on our end.  Make sure to note the date stamp for the
>various platforms in case the checks happened before your last update to
>your package.  Packages need to pass that check w/o errors and (with rare
>exceptions) warnings - so if you're seeing something on our end that
>doesn't jive with what you get on your end, please let me know.

Is there is mis-match between the pc and sun checks? For the limma package, 
the pc check seems to be on an older version of the package, one replaced 4 
or 5 days before the date stamp on the check. There are quite a few 
packages which are generating errors or warnings for the pc check but not 
the sun, so could there be a general problem of some sort?

Gordon

>Another tidbit, pointed out to me by Robert ... people who are seeing slow
>performance in their packages and are making heavy use of split() should
>try this out.  Apparently running split() on a named vector is much slower
>then one w/o names, so try removing the names from your vector before
>issuing your split() commands and see if that improves performance.
>
>Thanks
>-J