[PATCH] Add %includeifexists, a new type of hgrc include

[Matt Mackall]

On Tue, 2011-05-31 at 17:57 +0200, Martin Geisler wrote:
> Matt Mackall <mpm at selenic.com> writes:
> 
> > On Tue, 2011-05-31 at 17:30 +0200, Martin Geisler wrote:
> >> Matt Mackall <mpm at selenic.com> writes:
> >> 
> >> > On Tue, 2011-05-31 at 01:19 +0100, Tom Fitzhenry wrote:
> >> >> # HG changeset patch
> >> >> # User Tom Fitzhenry <tom at tom-fitzhenry.me.uk>
> >> >> # Date 1306801124 -3600
> >> >> # Node ID 029faced02ef51b7de5c6603f81580267d669c1e
> >> >> # Parent  f6a433671c06ea7986e0cf2c54686d1af7f22ba1
> >> >> Add %includeifexists, a new type of hgrc include
> >> >> 
> >> >> Use case: a user has a hgrc which is shared across multiple machines,
> >> >> /some/ of which have configuration specific to that machine. %include
> >> >> allows the user to include another configuration file, but causes a
> >> >> parse error if the included file does not exist.
> >> >
> >> > Perhaps we should just drop the parse error. This might make it harder
> >> > to debug problems with specifying the filename, but would otherwise be
> >> > harmless. Thoughts?
> >> 
> >> I would like that, along with a ui.debug message. Maybe we could then
> >> let showconfig output a warning so that it's very clear what's going on.
> >
> > Unfortunately, there are no ui objects at this layer.
> 
> Then I prefer just ignoring the exception -- I prefer that over
> introducing a new %tryinclude directive.

Ok, I've queued a fix that ignores include errors for nonexistent files.

-- 
Mathematics is the supreme nostalgia of our time.