Commit Graph

13 Commits

Author SHA1 Message Date
Jordan Hrycaj 3078c207ca
Aristo db implement distributed backend access (#1688)
* Fix hashing algorithm

why:
  Particular case where a sub-tree is on the backend, linked by an
  Extension vertex to the top level.

* Update backend verification to report `dirty` top layer

* Implement distributed merge of backend filters

* Implement distributed backend access management

details:
  Implemented and tested as described in chapter 5 of the `README.md`
  file.
2023-08-17 14:42:01 +01:00
Jordan Hrycaj 01fe172738
Aristo db integrate hashify into tx (#1679)
* Renamed type `NoneBackendRef` => `VoidBackendRef`

* Clarify names: `BE=filter+backend` and `UBE=backend (unfiltered)`

why:
  Most functions used full names as `getVtxUnfilteredBackend()` or
  `getKeyBackend()`. After defining abbreviations (and its meaning) it
   seems easier to use `getVtxUBE()` and `getKeyBE()`.

* Integrate `hashify()` process into transaction logic

why:
  Is now transparent unless explicitly controlled.

details:
  Cache changes imply setting a `dirty` flag which in turn triggers
  `hashify()` processing in transaction and `pack()` directives.

* Removed `aristo_tx.exec()` directive

why:
  Inconsistent implementation, functionality will be provided with a
  different paradigm.
2023-08-11 18:23:57 +01:00
Jordan Hrycaj 09fabd04eb
Aristo db use filter betw backend and tx cache (#1678)
* Provide deep copy for each transaction layer

why:
  Localising changes. Selective deep copy was just overlooked.

* Generalise vertex ID generator state reorg function `vidReorg()`

why:
  makes it somewhat easier to handle when saving layers.

* Provide dummy back end descriptor `NoneBackendRef`

* Optional read-only filter between backend and transaction cache

why:
  Some staging area for accumulating changes to the backend DB. This
  will eventually be an access layer for emulating a backend with
  multiple/historic state roots.

* Re-factor `persistent()` with filter between backend/tx-cache => `stow()`

why:
  The filter provides an abstraction from the physically stored data on
  disk. So, there can be several MPT instances using the same disk data
  with different state roots. Of course, all the MPT instances should
  not differ too much for practical reasons :).

TODO:
  Filter administration tools need to be provided.
2023-08-10 21:01:28 +01:00
Jordan Hrycaj 71c91e2280
Aristo db refactor tx paradim (#1674)
* Better error handling

why:
  Bail out on some error as early as possible before any changes.

* Implement `fetch()` as opposite of `merge()`

rationale:
  In the `Aristo` realm, the action named `fetch()` and `merge()` indicate
  leaf value related actions on the MPT, while actions `get()` and `put()`
   handle vertex or hash key related operations that constitute the MPT.

* Re-factor `merge()` prototypes

why:
  The most used variant of `merge()` should have the simplest prototype.

* Persistent DB constructor needs to import `aristo/aristo_init/persistent`

why:
  Most applications use memory DB anyway. This avoids linking `-lrocksdb`
  or any other back end libraries by default.

* Re-factor transaction module

why:
  Got the paradigm wrong. The transaction descriptor did replace the
  database one but should be handled separately.
2023-08-07 18:45:23 +01:00
Jordan Hrycaj 56d5c382d7
Aristo db traversal helpers (#1638)
* Misc fixes

detail:
* Fix de-serialisation for account leafs
* Update node recovery from unit tests

* Remove `LegacyAccount` from `PayloadRef` object

why:
  Legacy accounts use a hash key as storage root which is detrimental
  to the working of the Aristo database which uses a vertex ID.

* Dissolve `hashify_helper` into `aristo_utils` and `aristo_transcode`

why:
  Functions are of general interest so they should live in first level
  code files.

* Added left/right iterators over leaf nodes

* Some helper/wrapper functions that might be useful
2023-07-13 00:03:14 +01:00
Jordan Hrycaj 93a72025a1
Extended data Payload specs for the backend. (#1630)
why:
  For the main tree with root vertex ID 1, the leaf nodes hold the
  account data. These accounts may link to sub trees the storage root
  node ID of which must be registered here. There is no reverse key
  lookup on the backend.

note:
  These definitions are experimental. Also, there are some tests missing
  for validating Payload data conversions.
2023-07-05 21:27:48 +01:00
Jordan Hrycaj ccf639fc3c
Aristo db transaction based interface (#1628)
* Provide transaction based interface for standard operations

* Provide unit tests for new Aristo interface using transactions

details:
  These new tests combine and replace several single-purpose tests.
  The now unused test sources will be kept for a while to be eventually
  removed.
2023-07-05 14:50:11 +01:00
Jordan Hrycaj ff6673beac
Aristo db tidy up a bit (#1625)
* Slightly tighten some self-check conditions

* Redefined the database descriptor object as reference (to the object)

why:
  The upcoming transaction wrapper will work with a database reference
  rather than the object itself

* Append state before `save()` to the Aristo descriptor

why:
  This stae was previously returned by the function. Appending it to
  a field of the Aristo descriptor seems easier to handle.
2023-07-04 19:24:03 +01:00
Jordan Hrycaj dd1c8ed6f2
Aristo db update delete functionality (#1621)
* Fix missing branch checks in transcoder

why:
  Symmetry problem. `Blobify()` allowed for encoding degenerate branch
  vertices while `Deblobify()` rejected decoding wrongly encoded data.

* Update memory backend so that it rejects storing bogus vertices.

why:
  Error behaviour made similar to the rocks DB backend.

* Make sure that leaf vertex IDs are not repurposed

why:
  This makes it easier to record leaf node changes

* Update error return code for next()/right() traversal

why:
  Returning offending vertex ID (besides error code) helps debugging

* Update Merkle hasher for deleted nodes

why:
  Not implemented, yet

also:
  Provide cache & backend consistency check functions. This was
  partly re-implemented from `hashifyCheck()`

* Simplify some unit tests

* Fix delete function

why:
  Was conceptually wrong
2023-06-30 23:22:33 +01:00
Jordan Hrycaj 83dbe87159
Aristo db update foreground caching (#1605)
* Fix vertex ID generator state handling for rocksdb backend

why:
 * Key error in walk iterator
 * Needs to be loaded when opening the database

* Use non-zero sub-table prefixes for rocksdb

why:
  Handy for debugging

* Fix error code for missing key on rocksdb backend

why:
  Previously returned `VOID_HASH_KEY` rather than `GetKeyNotFound`

* Explicitly copy vertex data between internal table and function/result argument

why:
  Function argument or return reference may still refer to the same data
  object.

* Updated error symbols

why:
  Error symbol names for the hike module now start with the prefix `Hike`.

* Write back modified branch node into local top layer cache

why:
  With the backend available, the source of the branch node references
  might not be the top layer cache. So any change must be explicitely
  recorded.
2023-06-22 12:13:24 +01:00
Jordan Hrycaj 4b66f93274
Aristo db with storage backends (#1603)
* Generalised Aristo DB constructor for any type of backend

details:
  * Records to be deleted are represented as key-void (rather than
    key-value) pairs by the put-function arguments
  * Allow direct driver access, iterators as example implementation and
    for testing.

* Provide backend storage interface

details:
  Stores the top layer onto backend tables

* Implemented Rocks DB backend

details:
  Transaction based `put()` functionality
  Iterators (based on direct RocksDB access)
2023-06-20 14:26:25 +01:00
Jordan Hrycaj d7f40516a7
Detach from snap/sync declarations & definitions (#1601)
why:
  Tests and some basic components were originally borrowed from the
  snap/sync implementation. These have fully been re-implemented.
2023-06-12 19:16:03 +01:00
Jordan Hrycaj 0308dfac4f
Aristo db address sup trie items properly (#1600)
* Fix include

why:
  Eth67 not default yet so that got missed

* Rename `LeafKey` => `LeafTie`

why:
  Name is a pen picture of what this object is for. Also, it avoids the
  ubiquitous term `key`.

* Provided `getOrVoid()` wrapper for `getOrDefault()`

also:
  Provide `isValid()` syntactic sugar for `.isNil.not`, `!= 0` etc.
  Reorg descriptor source, split into sub-sources

* Bundled `NodeKey` objects with root ID and called it `HashLabel`

why:
  `NodeKey` (aka repurposed Hash265) objects are unique only within a
  particular sub-trie (e.g. storage slots) which are kept separated
  (i.e non-interleaved) by design. This is not applied to the backend
  as the map VertexID->NodeKey labelling the nodes needs not be injective.

  For the in-memory database (transaction) layers, the injective map
  VertexID->(VertexID,NodeKey) is used where the first field of the image
  tuple is the root ID of the sub-trie the `NodeKey` object is valid. So
  identical storage tries for different accounts can be represented.
2023-06-12 14:48:47 +01:00