[PATCH 3 of 3 RFC] chgserve: preimport enabled extensions

[Yuya Nishihara]

On Mon, 3 Oct 2016 07:32:47 +0100, Jun Wu wrote:
> This is the first 3 patches of the chg re-arch series. They may look simpler
> than expected at first sight, but locally I have 7 more patches, and the
> handling of "ui" is still not done yet.
> 
> I have chosen these 3 to start the discussion, as they reveal the plan to
> some extent, and seem to be most relevant to the next steps.
> 
>   - Should we hack "hg" / "dispatch" instead so they become aware of chg?
>     This leads to simpler code but since chg is experimental, I guess the
>     answer is no for now. But we may move chg to dispatch eventually.

If we can get rid of the entry-point script by the "dispatch" hack, I want to
go that way just because that "looks" cleaner than installing new script into
PATH.

>   - "chgserve" as an executable, will couple with "hgext/chgserver.py".
>     With the current direction, "chgserver" will make less sense as an
>     extension. I'm not sure if moving it out is a good idea or not.

Anyway I want to move chgserver.py to the core. Only reason I kept it as an
extension is I couldn't address circular imports between chgserver.py an
commandserver.py.