Posted: 07/03/08 12:17:24
I don't agree.
- Thousands files on hdd? Hmm.. this gives a good enhancement idea for DMD. Anyway, that's not a big deal, move on.
- Bigger lib? Who cares? Unused classes should be stripped from a target executable. (I know dmd can't do that for now, but gdc does, to my knowledge). This is not a Tango problem.
- Slower build times. No, if you don't use some modules, then these modules are not parsed and don't take any compilation time (except linking, maybe).
There are other minuses: it is yet another fork that only harms. Some people would ask about porting some stuff to MiniTango?, others would complain about it, etc. Someone should maintain it as well. Any volunteers? :)
But I do see one benefit out of it. There should exist a small subset of self-sufficient Tango library to ease porting DMD to other platforms. For ex., not everyone needs IO, net, collections etc. Dropping some non-essential features while focusing on essential ones would help to get up and running HelloWorld? applications faster. And such a subset does exist and is concentrated in a Tango.core package (please, correct me if I'm wrong).
I would also move Threads out of it (since they are not crucial for an application, unlike Mutexes - they are needed for synchronized statements), but it is just a matter of someones flavour, I think.