Formal Review Process

[Jesse Phillips]

On Wednesday, 12 June 2013 at 03:23:54 UTC, Jonathan M Davis 
wrote:
> If it's not pull-request ready, then it's not showing how it 
> would fit into Phobos.

I agree.

> Where do they fit into Phobos?

This has been what I'm challenging. The problem I'm trying to 
solve, and probably inappropriately doing so as part of the 
formal review, is as below.

1. We do not have a formal way to confirm a library should be put 
into Phobos
2. When Phobos does not provide the supporting modules we do not 
have a formal way to answer "Where do they fit in?"

I'm having a hard time concisely expressing myself with this. In 
principle I completely understand why you're taking this 
position, but practically I don't see it.

Let me go back to real examples: 
https://github.com/jacob-carlborg/orange

Orange has a directory for 'tests.' Phobos has generally not had 
unittests separate from the function/module being tested. Is that 
unacceptable? I don't know, there certainly was no objection when 
asked.

The repository provides a means to build the source into a 
library. Does it really make it harder to review because someone 
can build and use the code without checking out dmd, druntime, 
phobos and building themselves a testable Phobos library?

There is also a wiki folder, it has an orange ball. Ok, yes there 
are distractions.

I'll admit, I started from the docs, diving into the code as I 
needed and had I started from the github repository it wouldn't 
have been obvious where things would go in Phobos.