Eric Wilhelm wrote: > > A little here + a little there + a little more = bloat. No, it's not > optimal as-is, but I don't think that makes a good excuse to make it > less so. >
On reflection, getting past the 'lets have a --bloat-flag' and "Haven't got a clue" sillyness, I think you may well be right. My initial post came out of the fact that the only real world situation I have come across where a dependency on attributes.pm was missed by M::SD isn't fixed by "if threads::shared add attributes". My feeling is that this will add attributes.pm when it isn't needed and miss it when it is needed, at least as often as it turns out to be the right thing to do. So, I think the only realistic way you can cope with attributes by default is to add attributes.pm to your default list. However,there is an established method of adding known dependencies to M::SD. So I can see the reasoning for not doing anything specific for attributes.pm itself. IF it were the case that pp -o my.exe my.pl gave you a working exec in 99.9% of cases, then the case for including attributes.pm by default would be stronger, I think. But in the real world, you will almost certainly have to read the pp POD and understand how to include modules and libs, and work out when pp misses the odd module. As Steffen says, it is a developer tool. So, I'll turncoat and switch my vote for the 'do nothing' option. Regards Mark
