I should summarize what I've learned after my more recent post on this:
With 'long double', it is extremely expensive to keep "track of"
non-finites (NA, NaN and Inf) when doing basic arithmetics. The most
extreme cost is seen when the first value is non-finite, because there
is penalty added in
Sorry, "require" was the wrong word - "suggest" is better. I actually
didn't realize when I first wrote a package that I should create a
NEWS file because I was following the `package.skeleton()` route and
that wasn't something that was generated automatically. I suspect
other new package authors m
Mark,
As NEWS is not (yet) a required file for an R package on CRAN, we might do
well to encourage developers to provide that information. If moving with
the times means that more R package authors do that then I'll all for
allowing markdown. Many package authors do include markdown material
becau
On 05/06/2015 8:16 AM, Thomas J. Leeper wrote:
It's disappointing that many packages do not have a NEWS file. Perhaps
CRAN should require NEWS or CHANGELOG, as long as the system is being
reformed to potentially accommodate markdown anyway.
I think we should all be careful about suggesting work
> On 05-06-2015, at 14:16, Thomas J. Leeper wrote:
>
> It's disappointing that many packages do not have a NEWS file. Perhaps
> CRAN should require NEWS or CHANGELOG, as long as the system is being
> reformed to potentially accommodate markdown anyway.
>
I don’t really care about markdown.
But
It's disappointing that many packages do not have a NEWS file. Perhaps
CRAN should require NEWS or CHANGELOG, as long as the system is being
reformed to potentially accommodate markdown anyway.
-Thomas
Thomas J. Leeper
http://www.thomasleeper.com
On Fri, Jun 5, 2015 at 12:00 PM, wrote:
> Date
