I use hashdist outside of Sage on a daily basis, and its promise of "never 
compile anything twice" really works. It does bulletproof incremental 
builds, and the dependency tracking works very well in practice. 

If it weren't for distributing binary builds of Sage we should be using 
hashdist right now. In hashdist, the "installed package" is a bunch of 
symlinks back to the build store under ~/.hashdist/. This is why 
reinstallation is basically instantaneous, if the build is cached only a 
couple of symlinks are recreated. But you can't just tar up the install 
directory since its just a bunch of symlinks. You can of course copy files 
from the build store but they will keep their hardcoded paths and not work 
on a different computer.

One solution would be to export the hashdist build stores as conda packages 
to solve binary distribution. It just means that, internally, the build 
store would use long directory names as padding. There is a writeup at 
https://github.com/hashdist/hashdist/wiki/HDEP-1:-Crating-System-for-Redistributing-Packages,
 
but its not implemented yet.




On Friday, March 11, 2016 at 10:42:01 PM UTC+1, William wrote:
>
> Any chance you have a moment to make some remarks about hashdist in 
> this thread?   https://github.com/hashdist/hashdist 
>
>

-- 
You received this message because you are subscribed to the Google Groups 
"sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.

Reply via email to