add notes on datatype changes throughout the extension code

Author: ghazi-git

README.md

@@ -34,7 +34,9 @@ Chrome extension that adds a DevTools panel to manage IndexedDB data.
   likely contain the same properties for each key, making displaying the data in a table viable.
 - Only the following datatypes for table columns are supported: string, number, bigint, boolean, date and JSON. Other
   datatypes like sets, maps, typed arrays, ... are not supported, and the corresponding table columns will be marked as
-  such and hidden by default.
+  such and hidden by default. Also, once a column has a specific datatype, values that do not match that datatype will
+  be displayed as `undefined`. That means the extension is not the best for dealing with the same property storing
+  different datatypes (e.g. mixing numbers and dates).
 - The extension is designed to display data for small to medium-sized object stores. Data load will be slower for large
   object stores with hundreds of thousands of objects. You can still limit the number of objects to load from the "Table
   Settings" popover.
@@ -88,6 +90,44 @@ So, here I am creating this extension with the hope it may help others as it is
   - On the extension side, once the operation is triggered, the extension will start polling a specific property on the
     window object of the inspected page to check for the operation results.
   - The polling will stop once the results are found or after a set timeout.
+- Keeping track of possible datatypes handled by different parts of the code can get hard. So, here's a reference in the
+  hope of simplifying/improving it in the future or to help with adding new datatypes. This reference will highlight
+  the possible datatypes from getting data from an object store in the inspected page to displaying it in the table in
+  the extension's devtools panel:
+  - [inspected page] Get data from the object store: the datatypes here correspond to anything supported by IndexedDB
+    (i.e. any type supported by `structuredClone`)
+  - [inspected page] Auto-detect the supported datatypes:
+    - primitive types: string, number, boolean, bigint
+    - date objects
+    - timestamp: added datatype to represent integers that are milliseconds since epoch. Values are later formatted as
+      dates to make them easier to read. A value is auto-detected as timestamp if it's an integer greater than or equal
+      to `631152000000` (Jan 1st, 1990).
+    - JSON data: values are JSON-compliant objects or arrays.
+  - [inspected page] A datatype is auto-detected when 80% of the non-nullish values in the first 100 objects are of that
+    type. Otherwise, default to the "Unsupported" datatype (all values will be set as `undefined`)
+  - [inspected page] Passing the data between the inspected page and the extension using `eval` requires the data to be
+    JSON-compliant. So, date objects are converted to ISO-formatted strings and bigints to strings.
+  - [extension] Data received from the inspected page using `eval` and passed as is to AG-Grid's `rowData`.
+  - [extension] The `valueGetter` of each table column converts the value from `rowData` to the one that the table uses
+    for sorting and per-column filtering:
+    - for date columns, ISO-formatted date strings are converted back to date objects
+    - for bigint columns, bigint strings are converted back to bigint objects
+    - for timestamp columns, integers are converted to date objects
+    - for JSON columns, JSON objects/arrays are converted to JSON strings
+    - for any datatype, values that do not correspond to the datatype in question are set as `undefined`
+  - [extension] The `valueFormatter` converts what's returned by the `valueGetter` to strings that are displayed to the
+    user.
+  - [extension] `getQuickFilterText` converts what's returned by the `valueGetter` to strings used during the full table
+    search. It returns the same value as `valueFormatter`.
+  - [extension] During the cell value update flow, the cell editor parses, validates and converts the values to match
+    the column datatype. The `onCellEditRequest` first saves the value to IndexedDB, and if that's successful, saves the
+    value to `rowData` afterward.
+  - [extension] Saving the updated value to IndexedDB includes converting the value to a JSON-compliant value to be
+    passed to the inspected page (date objects to date strings, bigint objects to bigint strings, date objects from
+    timestamp columns to integers, JSON strings to JSON objects). Next, the inspected page will receive that value and
+    convert it to the original datatype for storage in IndexedDB (date strings to date objects and bigint strings to
+    bigint objects). After successfully storing the value, the extension will store the JSON-compliant value in
+    `rowData`.
 
 ## Running the project locally