[BioC] annotate incompatible with RdbiPgSQL !?!

[James W. MacDonald]

Hi Philip,

Philip Lijnzaad wrote:
> Dear all,
> 
> I took a while to 'solve' the following problem. I need to use both
> RdbiPgSQL and geneplotter. The latter loads, directly or indirectly,
> all of
> 
> annotate AnnotationDbi DBI RSQLite xtable lattice
> 
> I used to have an old version of annotate that I didn't bother
> upgrading and worked fine.  I recently did however, resulting in the
> indirect loading of DBI. This, in turn, is  incompatible with Rdbi,
> resulting in connection errors:
> 
> 
> Error in function (classes, fdef, mtable)  : unable to find an
> inherited method for function "dbConnect", for   signature "PgSQL"
> 
> I solved it by first library()-ing DBI, then Rdbi  and RdbiPgSQL.
> However, I don't think this  is very satisfactory. The ordering of
> libary() statements should be irrelevant and secondly, if I would
> just like to plot a few genes, it is overkill to also load complete
> database connection code. Would  a dependency "Suggests: DBI etc." ,
> rather than "Depends: DBI etc." not be more than enough for the
> annotate package?

No. Since BioC2.3 all annotation packages are based on RSQLite 
databases, so annotate requires DBI to be loaded in order to work correctly.

You are correct that this is an unsatisfactory situation, and namespaces 
are supposed to alleviate function masking, but only if they are set up 
correctly. So you should be contacting the maintainer of RdbiPgSQL 
directly (Hi Jianhua!) and asking him to fix his package.

HOWEVER, you are using a dirt-old version of R and BioC, and Jianhua may 
have fixed this problem many months ago, so you need to upgrade to 
R-2.9.0 and BioC2.4 first, to ensure this problem still exists.

Best,

Jim


> 
> Kind regards,
> 
> Philip
> 

-- 
James W. MacDonald, M.S.
Biostatistician
Douglas Lab
University of Michigan
Department of Human Genetics
5912 Buhl
1241 E. Catherine St.
Ann Arbor MI 48109-5618
734-615-7826