[darcs-devel] darcs patch: Implement gzReadFile (and 19 more)

[Ian Lynagh]

On Tue, Apr 19, 2005 at 04:03:10PM +0200, Tomasz Zielonka wrote:
> On Tue, Apr 19, 2005 at 09:03:56AM -0400, David Roundy wrote:
> > > BTW, I was going to try using Data.Map in Slurpy some day. I introduced
> > > get_slurp_context* functions to make the transition easier. I can work
> > > on this now.
> > 
> > Ian, what do you think of this? Should it make a difference?

I suspect it'll be worse if anything for "darcs whatsnew".

What might help for the whatsnew case is keeping a sorted list of
filenames in the repo. When we're not looking for adds we could use the
same list for looking in the working directory.

> Besides having directory contents always sorted, it could speed up many
> operations on Slurpy which require looking up files by their filepaths.

It might help in such cases, yes, especially if you have many files in a
directory.

This would conflict with having a sorted list for whatsnew, though.
Perhaps slurpies could be parameterised by how they represent the data.
I think we'd want to see some numbers showing a significant (in terms of
percentage time, I think) improvement before increasing the complexity,
though.


Thanks
Ian