fixup! Add SOP for schema changes (DataBiosphere/azul#2884)

Removed schema change embargo from spec and added reference to SOP.

Author: hannes-ucsc

docs/dcp2_system_design.rst

@@ -37,9 +37,13 @@ EBI Ingest (the primary channel for incorporating projects into the DCP/2), an
 adapter for processing analysis (meta)data from Terra workspaces, and adapters
 for high-level matrix data from a range of sources.
 
-All metadata is in JSON format and complies with the HCA Metadata Schema.
-Aside from minor schema changes that were necessary for processing the staging
-areas, the evolution of the schema is currently on hold.
+All metadata is in JSON format and complies with the `HCA Metadata Schema`_.
+Changes to that schema are made according to standard `DCP/2 operating
+procedures`_.
+
+.. _HCA Metadata Schema: https://github.com/HumanCellAtlas/metadata-schema
+
+.. _DCP/2 operating procedures: dcp2_operating_procedures.rst
 
 The DCP/2 only contains public (meta)data (not controlled access).
 
@@ -309,7 +313,7 @@ follows:
         UUID. Pick the row with the highest version.
 
     ii. read the ``inputs``, ``outputs`` and ``protocols`` properties (they're
-        all lists). 
+        all lists).
 
         For each input, output and protocol, extract the schema type and
         entity ID. Query the TDR table that corresponds to the schema type and
@@ -453,7 +457,7 @@ mapping between the two. Similarly, instead of allocating a random UUIDv4 for
 the descriptor ``file_id`` one could also derive a UUIDv5 from the SHA-1 or
 SHA-256 hashes of the data file's content.
 
-.. [#] 
+.. [#]
    If a file is referenced by multiple bundles using different file names, the
    DSS adapter stages multiple objects with the same content. This case occurs
    in the wild, but is of negligible impact (< 1% in volume, zarr store
@@ -503,7 +507,7 @@ files to an ``analysis_process`` in the ``links`` table (`metadata-schema
 Naming datasets and snapshots
 -----------------------------
 
-|nn| 
+|nn|
 
 This section contains specific details that anticipate that the DCP/2
 will soon need to support multiple snapshots of per catalog, at least one per
@@ -528,7 +532,7 @@ snapshots:
    labelling, sorting and filtering are available when listing datasets and
    snapshots using the TDR API. Additionally, IDs are hard to read to the
    human eye, and hard to distinguish visually, so as long as we manually
-   confer them between teams, names are preferred. 
+   confer them between teams, names are preferred.
 
 |ne|
 
@@ -655,7 +659,7 @@ descriptors, one for metadata files and one for ``links.json`` files.
 
   where
 
-  ``entity_type``  
+  ``entity_type``
     is the `HCA schema entity type`_ such as ``cell_suspension``.
 
   ``entity_id``
@@ -695,7 +699,7 @@ descriptors, one for metadata files and one for ``links.json`` files.
 
   where
 
-  ``file_name``  
+  ``file_name``
     is the ``file_name`` property from the file descriptor object for this
     data file.
 
@@ -705,7 +709,7 @@ descriptors, one for metadata files and one for ``links.json`` files.
 
   where
 
-  ``links_id`` 
+  ``links_id``
     is a UUID that uniquely identifies the subgraph. The DSS adapter uses the
     bundle UUID.
 
@@ -1061,19 +1065,19 @@ EBNF/Regex, starting at the ``strata`` non-terminal::
     strata = "" | stratum , { "\n" , stratum }
 
     stratum = point , { ";" , point }
-    
+
     point = dimension , "=" , values
-    
+
     dimension = "genusSpecies" | "organ" | "developmentStage" | "libraryConstructionApproach"
-    
+
     values = value , { "," , value }
-    
+
     value = [^\n;=,]+
 
 Examples:
 
-- Not stratified:: 
-  
+- Not stratified::
+
     ""
 
 - Stratified::
@@ -1231,8 +1235,8 @@ information about CGMs.
   CGM in the deprecated mechanism (`Describing CGMs as supplementary files`_)
 
 - the ``analysis_protocol`` contains an optional ``matrix`` module schema
-  containing the properties ``data_normalization_methods`` and 
-  ``derivation_process`` 
+  containing the properties ``data_normalization_methods`` and
+  ``derivation_process``
 
 Traversing the approximate CGM subgraphs, the Azul indexer infers a
 stratification tree of exactly the same structure as the one it derives from
@@ -1242,7 +1246,7 @@ mechanism (`Describing CGMs as supplementary files`_). The Data Browser
 exposes that tree in the same manner on the project details page. The inferral
 algorithm is identical to the one used for ``DCP/2-generated matrices`` with
 the one distinction that the subgraphs in the latter are exact, not
-approximate. 
+approximate.
 
 Additionally, the CGM analysis files are listed on the Files tab of the Data
 Browser.