Re: Commit (HEAD): UT_Map destructor virtual

From: Christian Biesinger (cbiesinger@web.de)
Date: Sat May 11 2002 - 10:55:07 EDT

  • Next message: Christian Biesinger: "Re: password prompt when choosing doc on rtf"

    On Sat, May 11, 2002 at 01:27:21PM +0200, Joaquín Cuenca Abela wrote:
    > On Fri, 2002-05-10 at 21:41, Christian Biesinger wrote:
    > > Make destructor of UT_Map virtual so that subclasses can override it and e.g. free the elements
    > > CVS: ----------------------------------------------------------------------
    > > CVS: Enter Log. Lines beginning with `CVS:' are removed automatically
    > > CVS:
    > > CVS: Committing in .
    > > CVS:
    > > CVS: Modified Files:
    > > CVS: ut_map.h
    > > CVS: ----------------------------------------------------------------------
    >
    > Please, remove this patch.
    >
    > UT_Map destructor should not be virtual (it was not virtual in the
    > beginning on purpose). Consider it a "final" class.

    Why that? Being able to override the constructor would make some things so
    much easier. Is the overhead of the vtable so bad?

    Would it be at least possible to add a possibility for UT_Map to
    automatically free its elements when it's destructed?

    > In the future, please consult any changes to UT_Map with me before
    > committing.

    Sorry for that; I consulted dom and thought that was sufficient.

    I made the destructor non-virtual again, though I'm still interested in
    your answer.

    -- 
    "They that can give up essential liberty to obtain a little temporary
    safety deserve neither liberty nor safety."
                                                     -- Benjamin Franklin
    




    This archive was generated by hypermail 2.1.4 : Sat May 11 2002 - 10:58:46 EDT