Version 0.2.4 released
Changelog:
- Fix root state update
- Remove
optionalfromjasondependency
drozdzynski
drozdzynski
Version 0.2.5 released
Changelog:
- Fix diff of
Datestruct - Rewrite tests from
HoundtoWallaby
1
drozdzynski
Version 0.3.0
Changelog:
- [BREAKING] Rename Cowbow handler module from
Storex.Socket.HandlertoStorex.Handler.Cowboy - Add support for Plug based apps
plug Storex.Plug - Update Storex application supervisor children spec
2
Sebb
drozdzynski
Version 0.4.0
- [BREAKING]
Storex.mutate/3is no longer based onsession_id - [BREAKING]
Store.init/2callback now need to return{:ok, state} | {:ok, state, key} | {:error, reason} - [BREAKING] Remove custom
Registrylogic - [BREAKING] Remove
connectioncallback from javascript client - New registry mechanism provides distributed mutations across the cluster
- Fix
terminatecallback inStorex.Handler.Plug - Added three callbacks to frontend client
onConnected,onErrorandonDisconnected
1
drozdzynski
Version 0.5.0
- [BREAKING] Frontend client fully rewritten
- [BREAKING] Updated the minimum required version of Elixir to
1.10 - Added support for non browser environment
1
drozdzynski
drozdzynski
Version 0.6.0
- [BREAKING] Keys of params in Store are now type
binaryinstead ofatom - Added message validation with structured casting
1
drozdzynski
drozdzynski
Version 0.7.0
- [BREAKING] Updated the minimum required version of Elixir to
1.16 - Fix type warnings emitted by
use Storex.Storeon Elixir 1.18 and newer - [BREAKING] A
FunctionClauseErrorraised inside a mutation that did match is no longer reported to the client asNo mutation matching .... Only the store’s ownmutation/5failing to match produces that error; anything else propagates with its original stacktrace, which stops the store process and the socket with it terminate/3is no longer skipped when the store module happens not to be loaded yet.Storex.Store.__terminate__/4gated the call onfunction_exported?/3, which answersfalsefor an unloaded module, so whether the callback ran depended on the code server rather than on the store- Housekeeping: dropped the unused
import Supervisor.Spec, warn: falsefromStorex.start/2(deprecated since Elixir 1.5, and thewarn: falsewas hiding it) and theregistry: Storex.Registry.ETSkey fromconfig/config.exs(that module was removed in 0.4.0 and nothing read the key). The project is now formatted, and CI checks it Storex.Registryreads (get_store/2,get_store_pid/2,get_store_instances/1,session_stores/1) now run in the calling process against the:protectedETS table instead of aGenServer.call. The registry process was a global serialisation point — every mutation performs at least one lookup — and the cost grew with the number of connections. Measured at 500 lookups per reader: 98.3ms against 16.2ms with 64 concurrent readers, 335.8ms against 56.2ms with 256. Writes and the:DOWNcleanup still go through the process- [BREAKING]
Storex.mutate/3andStorex.mutate/4return:ok. They used to return whatever the comprehension inStorex.PG.broadcast/1produced, which was the internal broadcast envelope repeated once per node ([broadcast: {:mutate, "Store", "reload", []}]) - Removed the unreachable
{:error, _}branch inStorex.PG.broadcast/1.:pg.get_members/2always returns a list; the error tuple was:pg2’s contract, and:pg2support went away in this release - The state a store reports on join is now read by asking the store process (
handle_call(:get_state, ...)on the generatedServer) instead of:sys.get_state/1, a debug function that was reaching into the process’s internal state shape from the outside on every join. If the process is gone by the time the join reads it, the client now gets an error frame instead of the connection exiting - A client could kill its connection process with a well-formed
errorframe:Storex.Message.cast/1accepted the shape butStorex.Socket.message_handle/2had no clause for it.errorframes only travel server to client, so the shape is no longer accepted and the frame is refused with1007like any other unknown type Storex.Handler.Plugclosed a malformed payload with a bare1007, passing the reason as the process exit reason instead of the close payload. It now sends1007with the reason, matchingStorex.Handler.Cowboy- Removed dead code:
Storex.Registry.session_pid/1(no callers, and its:ets.match/2pattern was a 4-tuple against 5-tuple records, so it never matched anything) andStorex.Handler.Cowboy.websocket_init/3(a cowboy 1.x callback, unreachable on cowboy 2.x) - [SECURITY] The SSR/HTTP path now runs the same store-name checks as the WebSocket path.
Storex.HTTPcarried its own copy of the resolution logic with onlyModule.safe_concat/1, missingCode.ensure_compiled/1and theStorex.Storebehaviour check, soGET /storex?store=Any.Module¶ms=%7B%7Dcalledinit/2on any loaded module whose name resolved and serialised the result back to the caller. Both transports now shareStorex.Store.resolve/1 - [BREAKING] [SECURITY]
Storex.Handler.Cowboyno longer accepts:binaryframes. It decoded them with:erlang.binary_to_term/1, which creates atoms out of bytes the client controls — a single 25-byte frame is enough — and passed the resulting term straight toStorex.Socket.message_handle/2, bypassing theStorex.Message.cast/1allowlist every other entry point goes through. Binary frames are now closed with1003 Storex.Handler.Plughad no clause for:binaryframes at all, so one raisedFunctionClauseErrorand closed the connection with1011. It now closes with1003like the cowboy handler- [SECURITY] A
mutationframe is now resolved against the session the server assigned to the connection, not thesessionfield carried by the frame. Any client could previously mutate — and read the resulting diff of — any other session’s store by naming its session id. Mutating other sessions on purpose is whatStorex.mutate/3andStorex.mutate/4are for - A
mutationfor a store the session has not joined now returns an error to the client instead of exiting the connection process - Store processes are now registered through a
Registrykeyed by{session, store}instead of being named:"#{session}_#{store}". Session ids are unique per connection, so the old naming created one permanent atom per session-store pair and could exhaust the atom table on a long-running node - Removed the
:pg2fallback, unreachable since OTP 24 - Updated dependencies
- Added
Storex.Test, a supported way to drive a store from a test with no socket and no browser.start_store/2,state/1,commit/3,broadcast/4andstop/1cover starting a store (including a store whoseinit/2refuses), reading its state, running a mutation and getting back the state, the diff and any reply, running aStorex.mutate/3fan-out and applying it, and stopping the store synchronously so a test can assert on whatterminate/3did. Until now this meant going throughStorex.SupervisorandStorex.Registry, both@moduledoc false, or driving a real browser through Wallaby. The diff is the reason it exists: it is the contract with the frontend and there was no other way to assert on it. A store started this way registers the calling process as its session and is stopped when the test ends
1