Brief article on benchmarks of Python repository with leading DVCSen

Nicholas Allen nicholas.allen at
Wed Feb 11 14:27:46 GMT 2009

Stephen J. Turnbull wrote:
> Nicholas Allen writes:
>  > It looks like they didn't use a shared repository in Bazaar so perhaps 
>  > the results are not that meaningful.
> Unfortunately, the feedback we (the authors of PEP 374) have so far
> from Python developers is to the contrary.  Anything that doesn't
> happen automatically is "too complex" to ask the Python developers to
> do, so whatever is most straightforward is what is going to benchmarked.
This is exactly why I think a global cache of downloaded revisions makes 
sense. Your web browser has a cache so why shouldn't Bazaar? It would be 
a trivial thing to code (I would imagine for someone familiar with the 
source) but would mean that the default use case would automatically be 
faster without the user having to setup a shared repository or learn 
anything else about Bazaar. I'm not sure if this would have sped up the 
benchmarks here as I think they were doing local branching  but it would 
certainly speed up remote branching in the case that the user didn't 
make a shared repository.

If you could configure the size of the cache you could make it have size 
0 to get the exact same behavior as now.


More information about the bazaar mailing list