Martin,
The old data set will certainly stay for a while (or indeed indefinitely,
why not). To make the transition as smooth as possible (I am aware that
"maps" is used in more places than I can ever check myself) I am including
the possibility of switching back to "legacy" mode using an environment
variable (or an extra function "world.legacy(TRUE)"), so that any code that
runs into trouble can be switched to the old database without needing to
edit the code itself.
Would it be a good idea to make the new maps package (beta) available
(e.g. via github) prior to submitting to CRAN? It is more or less ready.
Given the many packages that call maps, this could give maintainers a
chance to check things. For the moment, it's only the country borders &
names that have changed.
Alex
From: "Martin Maechler" <maechler at stat.math.ethz.ch>
I think R-sig-Geo is the best place to discuss future developments and
will be posting there soon.
I agree ... but the following is not about 'maps' in particular,
but package maintenance / compatibility ....
As Ray mentioned, I (with his help & feedback) have been preparing an
updated world map. I am going to be very cautious, but the many changes in
country names (post-1990 Europe!) mean that any package or code that uses
maps by calling map('world',region=...) to create a map using country
names, could be affected. Maintainers of packages calling "maps" may want
to contact me to make sure I don't break anything. In any case, I am
currently testing a beta version that includes the old data set as a fall
back.
I know these data sets are not very small, but I would still
strongly recommend that you
keep the old map, say as "world198x" (and change the "x" to a
digit if you know which one, or the "8x" to "7y" ... etc).
Storage and bandwidth have become considerably cheaper and so
the package could provide (almost) back compatibility ... and
also something interesting in itself (comparison of two very
similar maps with R; teaching, history, ...).
Best regards,
Martin
---
Dr. Alex Deckmyn e-mail: alex.deckmyn at meteo.be
Royal Meteorological Institute http://www.meteo.be
Ringlaan 3, 1180 Ukkel, Belgium tel. (32)(2)3730646
I have been somewhat surprised by the number of positive responses to my
message. I have found a candidate for the position of maintainer of the
maps, mapdata and mapproj packages. I have confidence that Alex Deckmyn
<alex.deckmyn at meteo.be> will do at least as good a job as I have done in
this regard.
The current (recent) releases of the packages still have my name as
maintainer, but the address specified points to my (now) old work
address which will autorespond to say I have now retired. However in
practise, I will still be able to read and respond to such emails for
the foreseeable future.
Among other improvements, Alex is working on setting up a more accurate
"world" database for maps and will submit this once it has been fully
tested.
Thanks everyone for your support.
Ray Brownrigg
[[alternative HTML version deleted]]