[Rd] ChangeLog for R packages {was "Wish list"}

Martin Maechler maechler at stat.math.ethz.ch
Wed Jan 21 13:56:58 MET 2004

>>>>> "Duncan" == Duncan Murdoch <dmurdoch at pair.com>
>>>>>     on Wed, 21 Jan 2004 07:18:25 -0500 writes:

    Duncan> On Wed, 21 Jan 2004 08:48:12 +0000 (GMT), you wrote:
    >> Encourage or force?
    >> Quite a few packages do already have a ChangeLog, and I
    >> took the thrust of this thread to be to require one or at
    >> least nag for one.

    Duncan> I think it should be required, and as with Rd files,
    Duncan> we'll check the form but not the content.  For
    Duncan> example, require an entry in a specified format for
    Duncan> the current version number, and have optional
    Duncan> entries for earlier version numbers.

Thomas mentioned a
Changelog: <filename>
entry in DESCRIPTION.  Using either 'NEWS' or (any
capitalization of) 'ChangeLog' for <filename> would allow for
both formats (see also below).

    Duncan> I have no idea what the Debian ChangeLog format
    Duncan> looks like (other than the R example!), but I'd say
    Duncan> if we do this, it would make sense to use the same
    Duncan> format as NEWS (and the Windows CHANGES file), or
    Duncan> put those two files into the new format.  They are
    Duncan> currently easy to edit manually and easy to print
    Duncan> and read.  I don't think the same could be said of
    Duncan> an XML format, but I could be wrong.

For some packages (where I don't use CVS), I'd definitely want
to stick with ChangeLog: It's more precise, and using Emacs
(other editors / IDE ?), creating entries is *very* convenient
and efficient (as said before: datetime, author, file name,
function name are all created automatically).  

Any volunteers for writing a "ChangeLog -> NEWS"
converter (in Perl most probably, since that's what we assume
for building anyway)?  If we had this in place, we could allow both
formats in the package sources, but install only the NEWS type
(and/or it's translations ...).


More information about the R-devel mailing list