Default view for a file should be its content

This should be pretty trivial really.

I have a branch which does some of this -- lp:~mwhudson/loggerhead/browse-tree -- but sadly it also tries to combine the current files/ and annotate/ urls into a single browse/ url and that bit is pretty horrible. But maybe it'll be useful to someone.

Okay, I've started to do some work on this--so far it doesn't look too hard, and in my brief testing on my local machine, it makes a reasonably-sized annotate take .3 seconds instead of 1.1 seconds, so I'd guess we'd save a significant percentage of the total CPU time being used on codebrowse by this.

I am also tempted to do what mwhudson did, and change the URLs at the same time, but for now I will focus just on making it not annotate.

I think there are other optimizations that can be done to the annotate controller after this bug, as well. (In particular, do we really have to get an entire inventory just so that we can get the file content and its name [not even its path]?)

A few thoughts.

Annotation is nice when wanted, lets make sure there is a way to do
that. Perhaps as a later iteration.

Inventories are not big - they are in 2a 2 UUIDS long (and thus
cheap). Iterating an entire inventory is a tree walk in the CHK store,
so can be slower. Looking up a patch is a path walk in the (parent,
basename) tuple hashmap for the inventory, so quite cheap.

-Rob

Hey Robert. Yes, of course, there will still be a way to do annotation, as a part of this bug.

When it comes to any other optimizations, I'm sure we can discuss them when the time comes.

I've fixed this. See the MP for more details.

One of the other advantages of this patch is that the annotate view now actually renders progressively, instead of doing all of the annotation and *then* rendering the page.