software/erp5: format json with ./format-json

It was formatted with a '\t', which json-schemas test does not allow.

software/erp5/instance-erp5-input-schema.json

@@ -480,11 +480,11 @@
     "wcfs": {
       "description": "Parameters for wendelin.core filesystem",
       "properties": {
-	"enable": {
-	  "description": "Whether to enable WCFS filesystem and use it to access ZBigArray/ZBigFile data. In WCFS mode wendelin.core clients (Zope/ERP5 processes) share in-RAM cache for in-ZODB data without duplicating it for every client. This cache sharing does not affect correctness as isolation property is continued to be provided to every client.",
-	  "default": false,
-	  "type": "boolean"
-	}
+        "enable": {
+          "description": "Whether to enable WCFS filesystem and use it to access ZBigArray/ZBigFile data. In WCFS mode wendelin.core clients (Zope/ERP5 processes) share in-RAM cache for in-ZODB data without duplicating it for every client. This cache sharing does not affect correctness as isolation property is continued to be provided to every client.",
+          "default": false,
+          "type": "boolean"
+        }
       }
     },
     "wendelin-core-zblk-fmt": {

[Kirill Smelkov]

Thanks, @jerome. I recall verifying this manually with jsonschema, but probably that test did not caught everything.

[Jérome Perrin]

Thanks @kirr We have an extra test in this test suite which also checks the indentation ( https://lab.nexedi.com/nexedi/slapos/blob/08615bc50b7de675bfeb2bf9850d195180f357fd/slapos/test/test_json_schema.py#L79-92 ). I guess using jsonschema only checked for the syntax

[Kirill Smelkov]

@jerome, thanks for explaining. And from that test I see we also have format-json which is useful to bring json/schema into canonical form (to which you already reffered in the commit title).