[erlang-questions] Systools and Release Handler

Tristan Sloughter t@REDACTED
Sun Jan 11 16:09:10 CET 2015


I'm hoping the OTP team can shed some light on what I consider an oddity
in systools and release_handler. I'm only bringing up one at this time,
it is the most often brought up confusion for users when using relx that
traces back to systools and release_handler.

systools make_script creates a .boot file with the name of the release.
When make_tar is called on the release that boot script is renamed
start.boot

release_handler relies on it being named start.boot

This means that a script for starting a node must check "does
RelName.boot exist?" to know which name to pass to -boot.

A small amount of code, but why bother ever having the boot script not
named start.boot?

There are some other confusing points when it comes to release_handler
but I'll save those for later (mostly because I haven't kept track).

But besides simply "why" I'm also wondering what the tools you use do to
handle this? Separate ways of starting in development than deployment?
Switches in scripts like I mentioned? And would modifications to
systools to always use one name or the other be accepted?

-- 
  Tristan Sloughter
  t@REDACTED



More information about the erlang-questions mailing list