Maybe consistency? If something is public, you provide everything needed for easy use and access. If it's not, you don't.
It is perhaps an unusual case, in that it's public elsewhere (i.e. anyone can grab the ksh93 source, and those interfaces may be treated as public on other OS's). OTOH, if the project team wants to wait awhile before making those interfaces public (to fully understand the consequences), then perhaps they should be willing to avoid tacitly encouraging their use until they're willing to make them public. This could of course be argued either way. I don't think it's worth slowing down the end result for. As such, anything like putting the lint libs out there that can be done later with very little extra trouble probably might as well be. I don't think there's going to be a huge rush of users wanting to add builtins. However, I think it might not be unreasonable if they were made available separately (as a plain tar file on the project page) _outside_ the tree, so that anyone wanting to play (with the full understanding that nothing they did would be accepted until the interfaces were no longer project private) could get a headstart. That way, if someone wants to start independent work on reconciling a less reduced set of builtins with the non-builtin versions, they could do that based on the (once it happens) integrated ksh93 rather than on their own separate build of it, which would increase the odds of their having something acceptable and/or simplify the merge work they'd have to do if and when the interfaces were made more than project private. Is that a workable compromise? This message posted from opensolaris.org
