destroy.test

Clone Tools
  • last updated 19 hours ago
Constraints
Constraints: committers
 
Constraints: files
Constraints: dates
nsf.c: - ignore in internall calls to "dealloc" protection settings - handle cyclical class dependencies during object system finalize - extend regression test

  1. … 2 more files in changeset.
nsfShadow.c - bump MethodEpoch when a tcl ::rename command happens on a nsf method (which might be cached in a Tcl_Obj) This fixes a crash reported by Arthur Schreiber nsf.c: - make NsfInstanceMethodEpochIncr() and NsfObjectMethodEpochIncr() accessible from all files using nsfInt.h - remove experimental code (backslash escaping for "," in parameter option parse

  1. … 4 more files in changeset.
- don't allow object creation to overwrite non-object cmds (e.g. procs)

  1. … 5 more files in changeset.
- deacitivate for the time being allowmethoddispatch (make it behave always like true) - added instead new flag "perobjectdispatch" to make behavior of ensembleobjects configurable. - The behavior for keepcallerself is currently only activated for the method-interface for object dispatch, since otherwise the following would be dangerous, since "o2 foo" would destroy o2 nx::Object create o1 nsf::object::property o1 keepcallerself true nx::Object create o2 { ::public method foo {} {o1 destroy} } o2 foo

  1. … 10 more files in changeset.
- nsf.c: start all error messages with a lower case word for consistency and to follow closer to Tcl's conventions

  1. … 14 more files in changeset.
- use alias-stubs for aliases pointing to objects. This allows us to distinguish between cases, where an object is dispatchable due to the alias or due to allowmethoddispatch (when the object happens to be a subobject and has therefore its cmd in the same namespace). The semantics are now: - aliases to objects are always dispatchable, no matter, how allowmethoddispatch is set. - direct subobjects of objects are currently on dispatchable when allowmethoddispatch is set.

  1. … 4 more files in changeset.
- alias reform: instead of calling first an alias with a stack frame followed by the dispatched of the aliased cmd, resolve aliases internally in dispatch. This has the advantage that we do not have to ignore the "transparent" stack frame in some sitations, and we have much less problems with the names of the aliased cmds (e.g. objects) in the introspection methods. Additionally, several problem cases disappeared from the regression test cases. In addition, the new approach is faster. - eliminating obsolete flag NSF_CSC_CALL_IS_TRANSPARENT

  1. … 9 more files in changeset.
- Added exerpimental object property "allowmethoddispatch" for child-objects to be called from parent objects via method interface. Background: both, per-object methods and childobjects are implemented via cmds in the same tcl namespace. Without special care, both are callable via the standard dispatch. Apparently, this is not always wanted. - handled allowmethoddispatch and keepcallerself in copy/move - set allowmethoddispatch per-default in XOTcl - removed visablilty of objects with "allowmethoddispatch" false in "info methods" and "info search methods" - extended regression test

  1. … 10 more files in changeset.
- removed TODOs from keepcallerself in destroy.test; calls were truely recursive, behavior was correct.

  1. … 1 more file in changeset.
- simplified usage of ObjectName() and ClassName() macros (no caller parenthesis needed) - added exerpimental object property keepcaller self (currently only evaluated by aliased objects)

  1. … 8 more files in changeset.
- tests/destroy.test: I added some tests on the robustness regarding unset traces being fired during the phyical OS delete phase. Checked with NSF_MEM_COUNT and valgrind.

nsf.c: - Reform of subclass/superclass handling. * unifying transitive subclass handling * localizing and documenting caching of subclass/superclass lists * eliminating repetitious computation of subclass lists * refactored code, renamed some functions to better refelect their purpose * improved documentation - fixed a potential crash for class deletion triggering implicit deletions further deletions referencing the parent class - extended regression test

  1. … 3 more files in changeset.
- reduce verbosity

- nsf.c: * integrated "-local" and fully qualified handling with ObjectDispatch to ensure proper behavior of mixins/next etc. * added "/obj/ -local ..." similar to "/obj/ -system ..." * added "nsf::object::disapatch /obj/ -local ..." similar to "/obj/ -local ..." * extended regression test (next from -local, fully qualified names, private methods, "...dispatch -local")

  1. … 8 more files in changeset.
- nx.tcl * renamed "attribute" to "property" * renamed "defaultAttributeCallProtection" to "defaultPropertyCallProtection" * renamed "nx::Attribute" to ""nx::VariableSlot" * renamed "BootStrapAttributeSlot" to "BootStrapVariableSlot" * renamed "createBootstrapAttributeSlots" to "createBootstrapVariableSlots" * removed method attributes * implemented old attributes definition in xotcl2 as method "parameter"

  1. … 32 more files in changeset.
- test.tcl: * don't export Test per-default * define Test as nx::Test * make Test parameter count 1 the default, change to higher numbers where needed

  1. … 15 more files in changeset.
- nsf: renamed nsf::isobject to nsf::object::exists - nsf: renamed nsf::qualify to nsf::object::qualify

  1. … 17 more files in changeset.
- changed named of method handling commands in nsf: ::nsf::methodproperty => ::nsf::method::property ::nsf::method => ::nsf::method::create ::nsf::methoddelete => ::nsf::method::delete ::nsf::alias => ::nsf::method::alias ::nsf::forward => ::nsf::method::forward ::nsf::setter => ::nsf::method::setter (updated regression test, docs, ...)

  1. … 24 more files in changeset.
- Altered Nsf_NextHashEntry() to re-init hSrchPtr when the number of expected entries differs from the number of real entries. This fixes a bug that Michael Aram detected, that happens when multiple hash buckets exist, but on deletion of an hash entries triggers some other deleted from the same hash table. - extended regression test.

  1. … 2 more files in changeset.
- check in correct version

- nsf.c: fixed a nasty bug within namespace deletion, when a deletion of one tcl cmd caused implicit deletions of other cmds in the same namespace. The classical idiom for looking over hash tables with Tcl_GetHashValue() and Tcl_NextHashEntry(hSrch) can lead to crashes (and has different behavior depending on the number of buckets). - added regression test

  1. … 3 more files in changeset.
- fixed scoping issue, when "-volatile" was used for object creation - added regression test for interaction between filters and function resolver (and volatile) - reactivated new volatile test in destroy.test - undone temporary fixes for volatile in serializer and nx.tcl

  1. … 5 more files in changeset.
- added deavtivated test for volatile via flag (unsetteled issue)

- changed CallStackGetActiveProcFrame() to return also CMETHD frames This allows to execute :volatile in a initcmd and to delete the object at its end. As a consequence, code like [CopyHandler new -volatile] copy [::nsf::self] $newName has to be changed to CopyHandler new { :copy [:uplevel ::nsf::self] [uplevel set newName] :destroy } - renamed CallStackUseActiveFrames() to CallStackUseActiveFrame() and ctx->framesSaved to ctx->frameSaved to reflect implementation

  1. … 5 more files in changeset.
- added regression tests for destroy-during-init

  1. … 1 more file in changeset.
- renamed nx regression tests .test to follow tcl conventions

  1. … 35 more files in changeset.