Blobs directory configuration (#549)

* Move configuration of blobs base path into client

* Rename variable in all places

* Update READMEs and fix 'its' spelling

* Add entry to CHANGELOG.md

* Add warnings when database is persisted but blobs are not

CHANGELOG.md

@@ -13,6 +13,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Add method to store for pruning document views [#491](https://github.com/p2panda/aquadoggo/pull/491)
 - Introduce `BlobStore` [#484](https://github.com/p2panda/aquadoggo/pull/484)
 - Task for automatic garbage collection of unused documents and views [#500](https://github.com/p2panda/aquadoggo/pull/500)
+- Blobs directory configuration [#549](https://github.com/p2panda/aquadoggo/pull/549)
 
 ### Changed
 

README.md

@@ -213,7 +213,7 @@ If you are not working with Rust you can create FFI bindings from the `aquadoggo
 
 As an application developer the interface to `aquadoggo` you are likely to use the most is the GraphQL query API. For whichever schema your node supports a custom query api is generated, you use this to fetch data into your app. Results from a collection query can be paginated, filtered.
 
-Fetch one "mushroom" by it's id, returning values for only the selected fields:
+Fetch one "mushroom" by its id, returning values for only the selected fields:
 ```graphql
 {
   mushroom: mushroom_0020c3accb0b0c8822ecc0309190e23de5f7f6c82f660ce08023a1d74e055a3d7c4d(

aquadoggo/Cargo.toml

@@ -68,7 +68,6 @@ sqlx = { version = "0.6.1", features = [
     "sqlite",
     "runtime-tokio-rustls",
 ] }
-tempfile = "3.7.0"
 thiserror = "1.0.39"
 tokio = { version = "1.28.2", features = [
     "macros",
@@ -112,5 +111,6 @@ rstest = "0.15.0"
 rstest_reuse = "0.3.0"
 serde_bytes = "0.11.12"
 serde_json = "1.0.85"
+tempfile = "3.7.0"
 tower = "0.4.13"
 tower-service = "0.3.2"

aquadoggo/src/config.rs

@@ -6,9 +6,6 @@ use p2panda_rs::schema::SchemaId;
 
 use crate::network::NetworkConfiguration;
 
-/// Blobs directory name.
-pub const BLOBS_DIR_NAME: &str = "blobs";
-
 /// Configuration object holding all important variables throughout the application.
 #[derive(Debug, Clone)]
 pub struct Configuration {
@@ -26,9 +23,6 @@ pub struct Configuration {
     /// _not_ recommended for production settings.
     pub allow_schema_ids: AllowList<SchemaId>,
 
-    /// Path to blobs directory.
-    pub blob_dir: Option<PathBuf>,
-
     /// URL / connection string to PostgreSQL or SQLite database.
     pub database_url: String,
 
@@ -44,6 +38,12 @@ pub struct Configuration {
     /// 2020.
     pub http_port: u16,
 
+    /// Path to folder where blobs (binary files) are kept and served from.
+    ///
+    /// **Warning**: When set to a temporary directory, make sure that also the database itself is
+    /// not persisted, otherwise you will run into data inconsistencies.
+    pub blobs_base_path: PathBuf,
+
     /// Number of concurrent workers which defines the maximum of materialization tasks which can
     /// be worked on simultaneously.
     ///
@@ -59,10 +59,10 @@ impl Default for Configuration {
     fn default() -> Self {
         Self {
             allow_schema_ids: AllowList::Wildcard,
-            blob_dir: None,
             database_url: "sqlite::memory:".into(),
             database_max_connections: 32,
             http_port: 2020,
+            blobs_base_path: PathBuf::new(),
             worker_pool_size: 16,
             network: NetworkConfiguration::default(),
         }

aquadoggo/src/db/stores/blob.rs

@@ -21,7 +21,7 @@ const BLOB_QUERY_PAGE_SIZE: u64 = 10;
 pub type BlobData = Vec<u8>;
 
 impl SqlStore {
-    /// Get the data for one blob from the store, identified by it's document id.
+    /// Get the data for one blob from the store, identified by its document id.
     pub async fn get_blob(&self, id: &DocumentId) -> Result<Option<BlobData>, BlobStoreError> {
         // Get the root blob document
         let blob_document = match self.get_document(id).await? {
@@ -36,7 +36,7 @@ impl SqlStore {
         document_to_blob_data(self, blob_document).await
     }
 
-    /// Get the data for one blob from the store, identified by it's document view id.
+    /// Get the data for one blob from the store, identified by its document view id.
     pub async fn get_blob_by_view_id(
         &self,
         view_id: &DocumentViewId,
@@ -62,7 +62,7 @@ impl SqlStore {
 
         // If there are no documents referring to the blob then we continue with the purge.
         if blob_reverse_relations.is_empty() {
-            // Collect the document view ids of all pieces this blob has ever referred to in it's
+            // Collect the document view ids of all pieces this blob has ever referred to in its
             // `pieces`
             let blob_piece_ids: Vec<String> = query_scalar(
                 "