[Rubygems-developers] More unhappy memory usage on bulk update

Mat Schaffer schapht at gmail.com
Sun Jun 1 10:20:01 EDT 2008


I'd like to bring back a discussion from http://rubyforge.org/pipermail/rubygems-developers/2007-May/002802.html

I noticed that rubygems is now using Marshal for what should be faster  
loading, but I think I may have stumbled upon some sort of bug in that  
system.   I edited the "say" statements in my rubygems to show whether  
they were coming from the site_ruby package or the 1.1.1 update gem.   
And also to output which index it was downloading and got this:

$ sudo gem install rails
Old Bulk updating Gem source index for: http://gems.rubyforge.org/
Loading Marshal.4.8.Z
Old Bulk updating Gem source index for: http://gems.rubyforge.org/
Loading Marshal.4.8
Loading yaml.Z

(Old == site_ruby installation)

Then I commented out Marshal.4.8, yaml.Z and yaml from the list of  
options and got this:
Old Bulk updating Gem source index for: http://gems.rubyforge.org/
Old Loading Marshal.4.8.Z
Old Bulk updating Gem source index for: http://gems.rubyforge.org/
ERROR:  While executing gem ... (Gem::RemoteSourceException)
     Error fetching remote gem cache: buffer error

This is after letting it run all night because I only have a 256M  
slice and it looks like it needs quite a bit more than that to complete.

Anyone have any thoughts as to why it appears to be restarting like  
this and not using the Marshal.4.8.Z download?  Or if there's an easy  
way to get more debug info?  And finally, have any of the developers  
considered some kind of solution that wouldn't require downloading the  
entire index?

Thanks,
Mat


More information about the Rubygems-developers mailing list