#16 seperate `fetch` and `emerge` for optimal resource usage
When using dialup connection, it is better to fetch the
packages in queue first and them emerge all fetched
packages (after disconnecting from internet)
This is very useful dialup users.
To implement this we can two Reader processes only for
fetch and one for emerge, after fetch processes fetches
a package the emerge process will emerge it, while
simultaneously the fetch process will start fethcing
next package in queue. This goes on.
There is one caveat in this idea though .. when
fetching some packages which are masked with
ACCEPT_KEYWORD="~arch" and if that package shares a
dependency with another non-masked package in the
queue, then two versions of the package (the
dependency) will be downloaded and installed -- this is
confusing and may screw up the system.
Logged In: YES
user_id=953915
Yes it will. Portage is undergoing a code rewrite so that
a proper API is available for programs such as porthole to
use. Once portage is ready and porthole is changed to use
the new API. It may be more readily doable.
Logged In: NO
I may be mistaken. But I think this feature was added to one
of the latest versions of portage. Correct me if I'm wrong.
Logged In: YES
user_id=953915
Portage now has file locking in place for multiple emerge
sesions, 1 for emerge -f, 1 for the actual emerge process.
I will implement a separate fetch setup in porthole when I
can. I currently have other things on the go at the moment.
Logged In: YES
user_id=980616
Will this really be needed? With version Portage 2.1_pre7 (i
think, maybe was 6) you can enable FEATURE="parallel-fetch"
witch will make portage start downloading all packages at
the same time as it starts the first emerge.
On the other hand, there ain't yet a nice way to see when
the parallel-fetch is completed, but that something to issue
with the devs of portage...