Re: pkglint and R package MASTER_SITES

[Roland Illig <roland.illig%gmx.de@localhost>]

On 09.08.2020 21:38, Brook Milligan wrote:
> I think the algorithm is:
>
> - does a package include math/R/Makefile.extension and does it
> override MASTER_SITES? if yes, then warn (or warn if
> PKG_DEVELOPER=yes).

Pkglint cannot know whether PKG_DEVELOPER is set or not.  Furthermore,
PKG_DEVELOPER is a misnamed variable.  It should rather be named
RUN_VERY_BASIC_SANITY_CHECKS and thus should be enabled by everyone, not
only pkgsrc developers.

What about this rule instead: Warn if an included file defines
MASTER_SITES and the including file overrides it with only a subset of
the default MASTER_SITES.

As far as I understood you, this would have caught the R packages.  The
rule sounds general enough to be useful for all other packages as well.

> If this had been in place before, pkglint would have caught early all
> the cases I have had to fix and would result in ignorable warnings
> for 2 of the 300+ R packages.

I don't like "ignorable warnings".

There are several places where the package author can suppress a pkglint
warning by writing a rationale comment near the location of the warning.
 For this scenario I am not sure whether there are any special keywords
to qualify as a rationale, therefore I would just regard any comment as
being a rationale.