[erlang-questions] Good reason for not using "releases"?
C. Florian Ebeling
Wed Oct 29 01:12:35 CET 2008
>>> Hello, I'm building my first erlang server program, and I've worked my
>>> way up from modules, to gen_server and gen_fsm, to supervisor, to
>>> application behavior.
>>> If this application is going to run on a linux host, and boot using a
>>> linux init script, does it make sense to take it to the next level of
>>> "release", or is that just more complication I don't need to deal with.
To me it seems like the erlang "application" concept is not quite
what you ususally would call an application. It encompasses code,
plus process(es) to be started an stopped in an orderly fashion.
But since processes are ubiquitous in erlang, you have to start something
in most cases. So an application feels to be analogous to a module
in other environments, and what is an application elsewhere is an
erlang release, because it integrates several applications, including
init order and such. So it is not as much over-engeneering as it
might appear in the beginning. Since you get upgrade facilities on
top you should really consider creating a system as an release.
More information about the erlang-questions