mirror of
https://github.com/hashicorp/packer.git
synced 2026-09-30 19:54:00 -04:00
Lucas Bajolet
479ca07324
When outputting the data from the provisioner so Packer can consume it, we are writing to a file. The only constraints on this file are that Packer manages its lifecycle, and that Packer core and the provisioner are synced-up in terms of the contents and the structure of the file. Since we are outputting the file, and its format, we can bundle the two together, and export them under a structure that both the provisioner and Packer core have access to, so we can then fill-in the blanks, and write a serialised version of this structure to disk. This is the approach taken in this commit. First we clean-up the existing code a bit, since some abstractions were a bit hasty, and did not necessarily make sense in all cases, so the code is consolidated under `Provision`, and we use that structure then to fill-in the relevant information before serialising it to JSON and writing it in the output file provided by Packer core.