Newsgroups: comp.lang.c++,comp.lang.smalltalk,comp.object
Path: cantaloupe.srv.cs.cmu.edu!rochester!cornellcs!travelers.mail.cornell.edu!news.tc.cornell.edu!news.cac.psu.edu!howland.reston.ans.net!ix.netcom.com!netcomsv!uu3news.netcom.com!medicus!johnm
From: johnm@medicus.com (John D. Mitchell)
Subject: Re: c++, smalltalk, and real-time
Message-ID: <1995Apr19.220415.9142@medicus.com>
Organization: Medicus Clinical Data Systems, Alameda, CA
References: <rshapiro-1404951247530001@esb.bbn.com> <3mp76p$ov3@stc06.ctd.ornl.gov> <1995Apr16.161344.4454@rcmcon.com>
Date: Wed, 19 Apr 1995 22:04:15 GMT
Lines: 20
Xref: glinda.oz.cs.cmu.edu comp.lang.c++:125108 comp.lang.smalltalk:23025 comp.object:30020

In article <1995Apr16.161344.4454@rcmcon.com> rmartin@rcmcon.com (Robert Martin) writes:
[...]
>Runtime speed in C++ is probably not affected by core OOP at all.
>Yes, virtual functions may require an extra microsecond or two to
>deploy.  (no more than that).  But that time is countered by the fact
>that the virtual deployment represents functionality that would
>otherwise have been addressed by if/else or switch statements.  

You are failing to recognize the issue of object construction and
destruction time.  I, as a user of an off the shelf class library,
cannot know how long a trivial looking piece of source code will
take (unless I happen to have the source and even then it can be
darn near impossible :-).  Sure this sort of speed issue can be
worked around but it's not necessarily easy to do in C++.

Also, all that object creation and destruction can certainly impact
speed just because they are doing so much heap memory allocation.

Take care,
	John
