hi,

Am Freitag, 5. April 2013, 19:46:32 schrieb Dirk Eddelbuettel:
> We cannot break exiting r-cran-* packages. We have 150+ of them.

i fully understand. and i never said you should break packages.

> Try to accomodate your needs within roxyPackage.

already did, it doesn't care now. but some packages might not build for others 
if not build with roxyPackage.

> | there are a couple of R plugin packages for RKWard[2,3] which we would
> | like to
> rwkward has been Debian package since 2005.

i know, i'm involved in its development since 2006. i'm talking about plugin 
packages *for* RKWard, which are all called "rk.<something>". that is, they 
all have a dot in their original name, so this issue affects all of them.

> | i see two possible *actual* solutions:
> |  - either change the policy to officially allow dots in package names,
> |  - or re-fix r-cran.mk and also fix all packages with invalid names.
> 
> The Debian R Policy does not matter.  You are referring to a 10-year _draft_
> document.

i figured there are debianized R packages, so the right procedure must be 
documented somehwere. i usually don't care how old a document is if i can't 
find a newer replacement. is there any reliable documentation on how to 
correctly debianize R packages, except for the draft? it's still referenced 
wherever you look for information on that, and all other aspects it mentions 
still seem to be valid, so it isn't obvious that it doesn't apply any longer.

so then, the code is law, and the draft is broken. rather than "dots in 
package names *should* be replaced by hyphens" it should actually read "dots 
in package names *must not* be replaced by hyphens", as that is the only way 
to come up with buildable packages.

again: i really don't care at all whether i must or must not replace dots in 
names, or whatever. but it *should* be documented in a way that you're able to 
understand the process. could you update the outdated draft so that it 
reflects what you're doing for some years now, and lift it from a draft to a 
valid guideline?


viele grüße :: m.eik

-- 
:. http://reaktanz.de/blog  ...  http://angstalt.de  ...  http://C3S.cc .:

Attachment: signature.asc
Description: This is a digitally signed message part.

Reply via email to