From mboxrd@z Thu Jan 1 00:00:00 1970 X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me X-Spam-Level: ** X-Spam-Status: No, score=2.6 required=5.0 tests=BAYES_20,INVALID_DATE, MSGID_SHORT,REPLYTO_WITHOUT_TO_CC autolearn=no autolearn_force=no version=3.4.4 Xref: utzoo comp.lang.ada:2827 comp.sw.components:382 comp.object:290 comp.software-eng:2264 comp.lang.modula2:1832 comp.lang.misc:3640 Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!ucsfcgl!cca.ucsf.edu!ucsfccb!dick From: dick@ucsfccb..ucsf.edu (Dick Karpinski) Newsgroups: comp.lang.ada,comp.sw.components,comp.object,comp.software-eng,comp.lang.modula2,comp.lang.misc Subject: Re: Inheritance & limited private types Message-ID: <2531@ucsfcca.ucsf.edu> Date: 27 Oct 89 00:04:07 GMT References: <48695@ricerca.UUCP> <6845@hubcap.clemson.edu> <126675@sun.Eng.Sun.COM> Sender: news@cca.ucsf.edu Reply-To: dick@ucsfccb.UUCP (Dick Karpinski) Organization: UCSF, Computer Center List-Id: In article <126675@sun.Eng.Sun.COM> grover@sun.UUCP (Vinod Grover) writes: >In article <6845@hubcap.clemson.edu> billwolf%hazel.cs.clemson.edu@hubcap.clemson.edu writes: >> It has long been recognized that the time to worry about product >> efficiency is AFTER the product has been developed and put through >> a profiler to determine where the bottlenecks are in the system, >> since in this way the high cost of maximizing efficiency can be >> directed to the points at which it will do the most good. > >For those of us without profilers, I suppose, there is no hope. Or perhaps >we should worry about efficiency before the product has been developed, or >perhaps not to worry about efficiency at all. It seems obvious to me that such of us that have that problem should fix it by making a profiler. Failing sufficient interest to do that job well, I have a simple technique for making inserted timing probes do the necessary work. My approach installs numbered probes into the application to collect timing information from the system and build a RAM table which is dumped in human readable form at end of run. In extreme cases (such as no clock) human keystrokes can be used to fake the clock. Remember, we only need crude information to detect which components are using most of the time. I'll happily tutor folks who need to do this by phone or email. Dick Dick Karpinski Manager of Minicomputer Services, UCSF Computer Center Domain: dick@cca.ucsf.edu (415) 476-4529 (11-7) BITNET: dick@ucsfcca or dick@ucsfvm (415) 658-6803 (Home) USPS: U-76 UCSF, San Francisco, CA 94143-0704 (415) 658-3797 (ans)