Yes, prepend your URL to the "binary_sites" key for the plist you're using. binary_sites = ( "http://my/cool/site/Roots/8A428", "http://darwinsource.opendarwin.org/Roots/8A428", ); .tar.gz is all that's supported right now. The main problem with supporting multiple archive formats is the network latency of checking 3x as many URLs, combined with validating if a .tar.bz2 is "newer" than the .tar.gz on the web server, and prefering it, or vice versa. Only dealing with one archive format means we only need to resolve "newness" between the darwinbuild cache in /usr/local/share/ darwinbuild/cache and self-built roots in Roots/, and we have full timestamp information. The first step for better root support is really supporting checksums in a better way, so we can tell if our local cache is stale. Then we'd need ways of prioritizing between multiple archive types, possibly using the mod times already sent in HTTP HEAD requests. Shantonu On Oct 6, 2005, at 2:20 AM, Sam Hart wrote:

Hi,

is there an easy way to get DarwinBuild to attempt to fetch roots from a separate location first before the default locations ?

I would like DarwinBuild to check my webservers for pre-built roots before defaulting to OpenDarwin's.

Also ... while I am here, are gzip'd n' tar'd files the only recognised file type for roots, or can I use bzip2'd n' tar'd files ... or maybe xar ?

S a m _______________________________________________ darwinbuild mailing list darwinbuild@opendarwin.org http://www.opendarwin.org/mailman/listinfo/darwinbuild