The clt thread at [L1 ] and Bug #1334947 [L2 ] revived the discussion of how to properly manage the reference count of Tcl_Objs.
This is an attempt at clarifying the issue, and a roadmap to improving both the core's refCount management as well as the documentation related to the subject.
1. Categorization
Joe English writes: There are roughly four classes of Tcl_Obj-related library routines:
- Constructors, which return a fresh Tcl_Obj with 0 refcount;
- Readers, which only read the value (but may cause shimmering).
- Consumers, which store a new reference to an existing Tcl_Obj, increment the refcount, and arrange to decrement the refcount at some unspecified point in the future.
- Mutators, which change the Tcl_Obj and can *only* be used with unshared Tcl_Objs (refcount = 0 or 1).
and Donal Fellows adds the category
- Hairy Monsters. Don't give these things refcount==0 objects since they will manipulate the refcount during their processing and might or might not retain a reference.
Please note that the same funtion may belong to different categories with respect to different arguments: for example, as currently implemented (up to Tcl8.5a4), the function Tcl_ObjSetVar2(interp, part1Ptr, part2Ptr, newValuePtr, flags) is a Reader wrt part1Ptr and part2Ptr, but a Hairy-Monster wrt newValuePtr. (DKF: It should be noted that there is no guarantee that Tcl_ObjSetVar2 will remain a reader wrt part2ptr; if we ever optimize internal memory management of arrays, that will likely change. This is why knowing which function is what with respect to each argument is hard.)
As a general rule, all Tcl commands should be considered to be Hairy-Monsters wrt the objects in the objv array.
We hope to improve the documentation wrt to the categorization of the different functions, and also to reduce significantly the population of Hairy-Monsters. As of today, Constructors and Mutators should be properly documented as such.
2. Rules for dealing safely with the different categories
Note first that Constructors are not an issue: there is no Tcl_Obj to manage before you call them.
The always-safe rules are:
- Mutators: pass an unshared object (refCount is 0 or 1). In order to respect copy-on-write semantics, make a copy for your use if you need to modify a shared object, and modify the copy.
- Readers, Consumers and Hairy-Monsters: Tcl_IncrRefCount(objPtr) before calling the library function, Tcl_DecrRefCount(objPtr) on return. This means: assume that every function in the Tcl library that is not a Mutator is a Hairy-Monster.
The optimal rules in terms of performance and code simplicity (but risky in light of incomplete documentation) are:
- Mutators: pass an unshared object (refCount is 0 or 1). In order to respect copy-on-write semantics, make a copy for your use if you need to modify a shared object, and modify the copy. Use Tcl_IsShared to determine whether you may modify, and Tcl_DuplicateObj to get the copy.
- Readers: if you pass an object with refCount==0, make sure to Tcl_DecrRefCount(objPtr) on return in order not to leak the object.
- Consumers: do not worry about reference counts as the consumer takes care of it, including the freeing of unneeded objects. This is fire-and-forget. Passing a fresh Tcl_Obj* to a consumer means you're through using it.
- Hairy-Monsters: Tcl_IncrRefCount(objPtr) before calling the library function, Tcl_DecrRefCount(objPtr) on return.
See also: