[erlang-questions] Erlang package manager

[Tristan Sloughter]

> The strongest reason I can see to build an Erlang centric PM is to
> support Erlang's existing deployment behaviors i.e. releases.

An Erlang centric PM should not do anything regarding releases.

The PM should focus on development and not for deployment. Releases
should be used to build target systems and deployed as such, maybe the
libs that are used to build the release/target system were fetched with
the PM at some point but the PM needs no knowledge of releases.

As for npm, gems, bundler, maven I know little about any of them. Except
that one is worth millions of dollars or something... And relies on
CouchDB and has hilarious outcomes when someone breaks semver rules.
Anyway... I don't think there is anything people are proposing we take
directly from those, instead to learn from prior art -- I'd suggest more
opam/cabal/cpan than those as useful prior art though.

Cargo, http://crates.io, is an example of the condensed knowledge taken
from those projects into a new package manager.

But I do like Maven's(and thus Leiningens) dependency handling, highest
spot on the dep tree wins. And we are using that for rebar3.

Tristan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://erlang.org/pipermail/erlang-questions/attachments/20150102/fd1f9a9e/attachment.htm>