Updated todo and PKGBUILD after 5.0.0 release.

Author: FedeDP

Extra/Arch/PKGBUILD

@@ -1,7 +1,7 @@
 # Maintainer: Federico Di Pierro <[email protected]>
 
 pkgname=libmodule
-pkgver=4.2.0
+pkgver=5.0.0
 pkgrel=1
 pkgdesc="C linux library to build simple and modular projects"
 arch=(any)
@@ -10,7 +10,7 @@ license=(MIT)
 depends=()
 makedepends=(git cmake)
 source=("${pkgname}-${pkgver}.tar.gz::${url}/archive/${pkgver}.tar.gz")
-sha256sums=("8bab8e23e324df7ffec6377a966ff71ced2c7de1bc59dfd806748fd9fcf5c51f")
+sha256sums=("fa82567c56313b69bea589f7a1a027217b722b9c885c2d645a826d8af307a6c9")
 
 prepare() {
     cd "${srcdir}/${pkgname}-${pkgver}"

TODO.md

@@ -1,109 +1,3 @@
-## 5.0.0
-
-### PubSub improvements
-- [x] avoid strdupping topic when creating pubsub_msg_t
-- [x] Only use one pubsub msg for every module, instead of  creating one for each recipient
-- [x] Keep a reference count for each "sent" pubsub msg: when it reaches 0 it means all recipients received the message, and then in can be freed
-- [x] Add an autofree field to module_tell/module_pubsub that automatically frees sent data when last recipient receives it
-- [x] Avoid strdupping module's name and ctx's name
-- [x] Avoid any allocation inside library and ALWAYS TRUST USER POINTER as core rule
-- [x] Add a self() alias to _self
-- [x] module_subscribe/unsubscribe should avoid checking if topic is registered in context
-- [x] Add new System messages: MODULE_STARTED/STOPPED
-- [x] Avoid telling system messages like MODULE_STARTED/TOPIC_REGISTERED to ourselves
-- [x] Test pubsub messagging for paused modules
-- [x] Fix memleaks!
-- [x] Unsubscribe shopuld not check if topic is registered in ctx as otherwise umsubscribing from a deregistered topic would not work.
-- [x] pubsub interface should take "const void *" instead of "const unsigned char *" as data
-- [x] Add a new parameter "bool global" to module_broadcast?
-- [x] FIX: if first context does not receive a global broadcast, message will be destroyed as it has 0 refs...
-- [x] Rename pubsub_msg to ps_msg inside msg_t
-- [x] Rename pubsub_msg_t to ps_msg_t
-- [x] Rename pubsub_priv_t to ps_priv_t
-- [x] Add a module_poisonpill() function, to send a system message to a module to stop it -> this will be enqueued in module's message queue, 
-conversely to module_stop that should stop module right away freeing all its enqueued messages.
-- [x] Always destroy messages in flush_pubsub_msg()? RIght now it delivers all of them if we're stopping looping on ctx
-- [x] module_poisonpill should only be sent to RUNNING modules
-- [x] when stop looping on a context, flush all pubsub messages to RUNNING modules only. Destroy messages for non-running modules.
-- [x] Drop (de)register_topic?
-- [x] mod->subscriptions map lazy creation
-- [x] Support regex topic subscriptions
-
-### Map
-- [x] FIx: avoid incrementing map size on value update
-- [x] Add test for update
-- [x] map->dtor should default to memhook._free, and fallback to default if map_set_dtor is called with NULL callback parameter 
-- [x] Add map_itr_t interface
-- [x] Add tests for new interface
-- [x] Improve hashmap implementation
-- [x] Add a stress test for hashmap implementation
-- [x] Fix map_clear implementation
-- [x] Add map_iterate test
-- [x] Set keydup and autofree in map_new()
-- [x] Drop map_set_dtor
-- [x] Destroy old value upon val update
-- [x] Avoid destroying map value on update with same pointer; just return MAP_OK
-
-### Stack
-- [x] stack->dtor should default to memhook._free, and fallback to default if stack_set_dtor is called with NULL callback parameter 
-- [X] Add stack_itr_t interface
-- [x] Add tests for new interface
-- [x] Stack_iterate to follow same logic as map_iterate
-- [x] Set autofree in stack_new()
-- [x] Drop stack_set_dtor
-
-### Queue
-- [x] Add queue API
-- [x] Add doc
-
-### Generic
-- [x] Add some diagnostic API, eg: module_dump() (to dump each module's state)
-- [x] Add a module_load/unload function, to load a module from a compiled object at runtime
-- [x] simplify interface for module_load/unload? ie: module_load(path, ctx_name)
-- [x] Add test.so build to Easy example
-- [x] Re-evaluate module_register/deregister parameters (ie: self_t should not be a const!)
-- [x] Provide a default (weak symbol) main() that just runs modules_ctx_loop() on any ctx it found
-- [x] Rename MODULE_DEFAULT_CTX and MODULE_MODULE_MAX_EVENTS to MODULES_*
-- [x] Actually call init() callback first time module is started, even without passing from evaluate_module (thus without looping ctx)
-- [x] module_load/unload should use RTLD_NOLOAD flag instead of yet another hashmap
-- [x] module_get_name/ctx to return a strdup string
-- [x] Rename modules_set_memalloc_hook to modules_set_memhook() + rename memalloc_hook to memhook_t
-- [x] stop() and start() should avoid err_str parameter.
-- [x] Stop looping on context when there are no RUNNING modules
-- [x] Cleanup some out of date comments in examples
-- [x] Avoid exposing in modules.h main() and modules_pre_start() functions: they're not part of libmodule's API and cannot be called as functions.
-- [x] Rename module to mod_t and m_context to ctx_t
-- [x] Rename module_poll_t to fd_priv_t
-- [x] FIX: avoid sending with MODULE_STOPPED pubsub message a not-exishtent reference to mod->self when deregistering, as module gets freed right after. Send NULL.
-- [x] FIX: when deregistering, remove module from context before stopping it
-- [x] Rename userhook to userhook_t, following other types
-
-### Doc
-- [x] module_dump
-- [x] module_subscribe/unsubscribe
-- [x] self()
-- [x] module_register/deregister
-- [x] module_load/unload
-- [x] module_tell/publish/broadcast
-- [x] MODULE_STARTED/MODULE_STOPPED new sysmessages
-- [x] Avoid telling system messages like MODULE_STARTED/TOPIC_REGISTERED to ourselves
-- [x] Document main() weak symbol!
-- [x] Add a new page about trusting pointers
-- [x] modules_set_memhook/memhook_t
-- [x] new stop behaviour (stop will always destroy pubsub messages instead of delivering them, except when a module gets stopped through module_poisonpill)
-- [x] module_poisonpill
-- [x] New modules_loop behaviour (stop looping when no RUNNING modules)
-- [x] New Map API
-- [x] New Stack API
-- [x] Dropped (de)register_topic!
- 
-### Samples
-- [x] Fix samples
-
-### Test
-- [x] Fix tests
-- [x] add module_dump tests
-
 ## 5.1.0
 
 ### Submodules