From 20430d450e166ec80688053cfd0d501a7a9cf4a8 Mon Sep 17 00:00:00 2001
From: Pierre Neyron <pierre.neyron@imag.fr>
Date: Thu, 6 Aug 2020 14:58:34 +0200
Subject: [PATCH] [spec] add README for the inputs and outputs

---
 spec/input/README.md  | 20 ++++++++++++++++++++
 spec/output/README.md |  5 +++++
 2 files changed, 25 insertions(+)
 create mode 100644 spec/input/README.md
 create mode 100644 spec/output/README.md

diff --git a/spec/input/README.md b/spec/input/README.md
new file mode 100644
index 0000000000..88f837d209
--- /dev/null
+++ b/spec/input/README.md
@@ -0,0 +1,20 @@
+How to update input files:
+==========================
+
+data_* files:
+-------------
+Uncomment the corresponding `gen_stub()` line in ../oar_properties2_spec.rb and re-run rspec.
+
+oar_* files:
+------------
+Except the oar_empty.json file which should not have to be regenerated, the other files must be forged from the output of the OAR API.
+
+Steps are:
+- fetch the OAR API output for the site: `curl 'https://api.grid5000.fr/stable/sites/nancy/internal/oarapi/resources/details.json?limit=999999' > oar_site.json`
+- sort the objects with jq: `cat oar_site.json | jq -S "." > new_oar_site.json; mv {new_,}oar_site.json`
+- replace the tested cluster names, e.g. with sed: `sed -ie 's/graffiti/clustera/g' oar_site.json`.
+- modify the oar_site.json file as needed for what should be tested.
+
+Since using `vimdiff` or `meld` on the files is very slow, one can use the `diff` and `patch` commands...
+
+Then verify using `git diff` that the generated data matches what you expect.
diff --git a/spec/output/README.md b/spec/output/README.md
new file mode 100644
index 0000000000..f79fbddefe
--- /dev/null
+++ b/spec/output/README.md
@@ -0,0 +1,5 @@
+How to update the output files:
+===============================
+
+To update the files, just remove them and run rspec again, they will be regenerated.
+Then verify using `git diff` that the generated data matches what you expect.
-- 
GitLab