Merge metaconfig units into the perl source

[jmdh]

This is a (my suggestion for) a long-term goal. When I asked why metaconfig.git was separate from the main source tree at the recent Perl 5 hackathon, the responses were along the lines of that it contained a lot of stuff that wasn't relevant to building or using perl.

Which is sort of true and sort of not, since metaconfig.git is an essential part of the process of maintaining perl. I think that, eventually, it should be merged back to perl.git, but I recognise that there is a way to go before this would be considered acceptable. I hope that creating an issue here is an acceptable way to record this goal.

As I understand it, metaconfig is designed to be run from the project source tree, and so to use it a the moment we need to create symlinks in both directions, and this complex environment setup is the main practical consequence of having it split out. This raises the barrier for use. My own driver for this is that it complicates the Debian packaging (which we have just updated to correctly regenerate configure to make sure we can modified it if we need to), and I hope we might be able to simplify this in the future.

Note: I'd like to treat this discussion separately from any discussion about whether a separate maintenance team or pumpkin for Configure is a good idea, as I believe that it would be perfectly fine to have this even if the metaconfig units were integrated. (I mention this only in case it comes up, not because I think that latter discussion is something that we need to have).

I think that this is something which should be considered once #27 and #28 are resolved.

[Tux]

My (only) problem with this is that it adds 1000+ files to the perl source tree, which nobody out of the configuration group is or will be interested in, slowing down grep and adding to confusion.
Please prove me wrong.

[jmdh]

On Tue, Oct 17, 2017 at 01:48:04PM +0000, H.Merijn Brand wrote: My (only) problem with this is that it adds 1000+ files to the perl source tree, which nobody out of the configuration group is or will be interested in, slowing down grep and adding to confusion. Please prove me wrong.

I do think that the most noticable effect will be more people who understand the configuration system, and not more confused people. As a percentage increase in size of the code base I don't think it would be significant, especially once the delta between upstream dist has been reduced a bit. It's definitely not something I'm suggesting happens now, but a nice ideal. Cheers, Dominic.