>>>>> "Dirk" == Dirk Eddelbuettel <[EMAIL PROTECTED]>
>>>>> on Mon, 19 Jan 2004 12:55:11 -0600 writes:
>> For CRAN to list changes to other packages would require cooperation
>> from all the package developers. If the maintainer of acepack isn't
>> sufficiently together to reply to your messages, he probably won't be
>> keeping up with other aspects of change tracking. Even trying to
>> extract a NEWS or Changelog file might not work -- eg for survival the
>> Changelog file is Terry Therneau's change log, not my log for changes to
>> the R port.
Dirk> Debian uses a convention that (somewhat simplified)
Dirk> if the package is a standard upstream package then
Dirk> its changelog or news or changes or ... file
Dirk> is installed as ChangeLog
Dirk> and the Debian maintainer changes go into ChangeLog.Debian
Dirk> else if the package is a 'native' Debian package
Dirk> the changelog is installed as ChangeLog
Dirk> fi
Dirk> R CMD check could warn if there is not ChangeLog or Changes file, and
Dirk> thereby suggest that authors provide them.
"Of course" "R CMD check" would have the harder job of also have
to work properly on non-Debian, non-Linux, even non-Unix systems..
For Windows, there's already the filename capitalization mess.
[Kurt?]
Dirk> R CMD install could install them in a standard place as we control the
Dirk> installation mechanism.
indeed.
Dirk> Would that help? In order to parse a Changelog, we'd
Dirk> need a common format. We could start with the
Dirk> standard emacs mode (with vi or, mercy on them,
Dirk> notepad users can emulate free-hand). Or is all this
Dirk> too restrictive?
I would be very much in favor of this. But then I'm biased:
Using Emacs 100% and using ChangeLog's for most of my packages
{for those who don't know: "C-x 4 a" = add-change-log-entry-other-window
this works even from an PKG/R/foo.R file, adding to toplevel
PKG/ChangeLog, automatically inserting the R/foo.R filename
(and a function name) there.
}
Having a common format for parsing would allow for quite a few
things, one of the first being a translation to HTML.
Dirk> The good thing is that enough rigour was put into CRAN
Dirk> practices early so ensure a very decent level of
Dirk> conformity across packages. We could try to extend this.
Indeed!
Honestly, I did have the same idea after Frank's post --- not
knowing the Debian setup --- but didn't post *because* of my
biases (above).
Martin Maechler <[EMAIL PROTECTED]> http://stat.ethz.ch/~maechler/
Seminar fuer Statistik, ETH-Zentrum LEO C16 Leonhardstr. 27
ETH (Federal Inst. Technology) 8092 Zurich SWITZERLAND
phone: x-41-1-632-3408 fax: ...-1228 <><
______________________________________________
[EMAIL PROTECTED] mailing list
https://www.stat.math.ethz.ch/mailman/listinfo/r-devel
