[Bioperl-l] Immediate-effect deprecations (was: bioperl-run and Bio::Root::AccessorMaker)

[Sendu Bala]

Alex Lancaster wrote:
> In packaging bioperl-run for Fedora, I think I stumbled across a bug
> in the bioperl-run package.  It appears from this edit:
> 
> http://code.open-bio.org/cgi/viewcvs.cgi/bioperl-live/Bio/Root/Attic/AccessorMaker.pm?hideattic=0&cvsroot=bioperl
> 
> that Bio::Root::AccessorMaker was removed in bioperl 1.5.x, but
> bioperl-run 1.5.2_100 still contains modules that use this module:
> 
> $ cd bioperl-run-1.5.2_100
> $ grep -r AccessorMaker  *
> Bio/Tools/Run/Phylo/Forester/SDI.pm:use Bio::Root::AccessorMaker (
> Bio/Tools/Run/JavaRunner.pm:use Bio::Root::AccessorMaker ('$'=>[qw(jar
> class min_version)]);
> Bio/Tools/Run/AbstractRunner.pm:use Bio::Root::AccessorMaker
> ('$'=>[qw(input_file output_file)]);

It looks like I've implemented a similar idea to AccessorMaker and 
AbstractRunner in Bio::Root::Root->_set_from_args() and 
Bio::Tools::Run::WrapperBase->_setparams(). Since nothing uses 
AbstractRunner I propose deprecating it immediately.

Forester::SDI and JavaRunner have no tests which is why we didn't notice 
the problem. Since they've been out of use for a number of years now I 
also propose their immediate deprecation. Alternatively, it may not be 
too difficult to just update them to use _set_from_args and _setparams, 
but I've nothing to test against (and JavaRunner is self-described as 
"probably incomplete").


I can remove the modules from cvs and create bioperl-run-1.5.2_101, 
resolving the packaging issue. I plan on doing precisely this within the 
next seven days unless someone puts a hand up to stop me.


[BCC: author, Juguang Xiao]