Re: version conflict for package "Tcl": TCL_VERSION vs TCL_PATCH_LEVEL

Donald G Porter <dgp@xxxxxxxx>:
Olaf Dietrich wrote:
Okay, so _without_ stubs, you recommend to use

Tcl_InitStubs(interp, TCL_VERSION, 1);

(works with all patchlevels of the MAJ.MIN version)
and _with_ stubs:

Tcl_InitStubs(interp, TCL_VERSION, 0);

No, I recommend the latter in both cases.

However, you were rejecting that. You were explicitly asking for a
formulation that would enforce a "no future minor release" constraint,
so I gave you the answer you were asking for.

If I remember correctly I found out in earlier tests that my extension
crashes when loaded in "future" minor releases, if built *without*
stubs support ("future" meaning newer than the one used for compilation.)

I don't expext this problem to be solved when compiling with
"Tcl_InitStubs(interp, TCL_VERSION, 0)" but *without* "-DUSE_TCL_STUBS".
However, I will do some more tests to find out ...

Olaf
.

Relevant Pages

  • Re: interpreter vs. compiled
    ... I note that IronPython and Python's pickle.dumps do not return the ... you'll reexecute stubs enough to outweigh the length of time ... Whereas, CPython can disregard ... The trade-off is compilation ...
    (comp.lang.python)
  • cannot compile with stubs - undefined symbols
    ... The compilation was like this: ... Linker gave me a lot of undefined Tcl_* functions. ... and how to compile a stand alone apps. ... I suceeded with building dinamic library using stubs nevertheless. ...
    (comp.lang.tcl)
  • Re: version conflict for package "Tcl": TCL_VERSION vs TCL_PATCH_LEVEL
    ... (works with all patchlevels of the MAJ.MIN version) ... and _with_ stubs: ... However, you were rejecting that. ...
    (comp.lang.tcl)