Morphic Splitters Team progress report

Avi Bryant avi.bryant at gmail.com
Wed Mar 16 12:01:03 UTC 2005


On Wed, 16 Mar 2005 08:26:32 +0100, Cees de Groot <cg at cdegroot.com> wrote:
> [periodical team report to squeak-dev]
> 
> We've been splitting like madmen - at the moment using Monticello, which
> works fine but is slowish (maybe better tooling/process will be the result
> of discussion in some of the other teams).

That's definitely a problem we're aware of - one of our goals with
Monticello 2.0 is to provide much better performance for large
packages.  I've been devoting what spare moments I have to doing
optimization work on MC2 (implementing an indexed binary file
repository, for example) while Colin pushes the main body of code
forward, and although it's not ready for use yet I think we're making
good progress.  One thing we haven't done yet, if someone wants to
help, is designed or implemented a networked repository.  The problem
is that, like ENVY, MC2 addresses individual method versions.  This is
part of what makes it fast - it usually only has to ask the repository
for a handful of changed methods - but it also means that in the worst
case, when loading a new package, it has to make hundreds of requests.
 This won't work very well with a naive HTTP or FTP implementation of
one file per method; even just using that approach in a local
directory proved to be unacceptably slow.  So if anyone has an
experiment they want to try here (p2p, cees?), that would be great.

Avi



More information about the Squeak-dev mailing list