sup talk - Jan 2008 - [PATCH] return sources in id-order, with non-archived sources first

To speed up the user''s view of polling, deal with sources that affect
the
inbox before (for example) large mailing lists. It seemed tidier to implement
this in Index rather than PollManager.

I changed "@sources.values" to "sources" in some other
methods for clarity,
which I think in this case outweighs lost efficiency.

---
 lib/sup/index.rb |   12 ++++++++----
 1 files changed, 8 insertions(+), 4 deletions(-)

diff --git a/lib/sup/index.rb b/lib/sup/index.rb
index 1997c92..b2a1612 100644
--- a/lib/sup/index.rb
+++ b/lib/sup/index.rb
@@ -125,9 +125,13 @@ EOS
     @sources[source.id] = source
   end
 
-  def source_for uri; @sources.values.find { |s| s.is_source_for? uri }; end
-  def usual_sources; @sources.values.find_all { |s| s.usual? }; end
-  def sources; @sources.values; end
+  def sources
+    ## favour the inbox by listing non-archived sources first
+    @sources.values.sort_by {|s| s.id }.partition {|s| !s.archived? }.flatten
+  end
+  
+  def source_for uri; sources.find { |s| s.is_source_for? uri }; end
+  def usual_sources; sources.find_all { |s| s.usual? }; end
 
   def load_index dir=File.join(@dir, "ferret")
     if File.exists? dir
@@ -512,7 +516,7 @@ protected
         File.chmod 0600, fn
         FileUtils.mv fn, bakfn, :force => true unless File.exists?(bakfn)
&& File.size(fn) == 0
       end
-      Redwood::save_yaml_obj @sources.values.sort_by { |s| s.id.to_i }, fn,
true
+      Redwood::save_yaml_obj sources.sort_by { |s| s.id.to_i }, fn, true
       File.chmod 0600, fn
     end
     @sources_dirty = false
-- 
1.5.3.7

Reformatted excerpts from Grant Hollingworth''s message of
2008-01-18:
> To speed up the user''s view of polling, deal with sources that
affect
> the inbox before (for example) large mailing lists. It seemed tidier
> to implement this in Index rather than PollManager.
Very nice. One minor comment below.
> I changed "@sources.values" to "sources" in some other
methods for
> clarity, which I think in this case outweighs lost efficiency.
Sounds good. And efficiency is the least of our concerns!
> +  def sources
> +    ## favour the inbox by listing non-archived sources first
> +    @sources.values.sort_by {|s| s.id }.partition {|s| !s.archived?
}.flatten
> +  end
Why bother sorting by id at all? Why not just something like:
  @sources.values.sort_by { |s| s.archived? ? 0 : 1 }


-- 
William <wmorgan-sup at masanjin.net>

* William Morgan [Fri Jan 25 00:27:14 -0500 2008]:
> > +  def sources
> > +    ## favour the inbox by listing non-archived sources first
> > +    @sources.values.sort_by {|s| s.id }.partition {|s| !s.archived?
}.flatten
> > +  end
> 
> Why bother sorting by id at all? Why not just something like:
>   @sources.values.sort_by { |s| s.archived? ? 0 : 1 }
I probably did that because it''s likely that an inbox is the first
source.
It''s not really necessary. I think partition is clearer than sort_by in
this
case.

How about changing @sources to an array?

Reformatted excerpts from Grant Hollingworth''s message of
2008-01-25:
> How about changing @sources to an array?
The id is important (it''s stored directly in the index), so I
don''t want
removing a source to change the id of all following sources.

I''ve merged this into next.

-- 
William <wmorgan-sup at masanjin.net>