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=1.1 required=5.0 tests=BAYES_05,REPLYTO_WITHOUT_TO_CC autolearn=no autolearn_force=no version=3.4.4 X-Google-Language: ENGLISH,ASCII-7-bit X-Google-Thread: f849b,5aa763fe62c20184 X-Google-Attributes: gidf849b,public X-Google-Thread: 115aec,5aa763fe62c20184 X-Google-Attributes: gid115aec,public X-Google-Thread: 103376,5aa763fe62c20184 X-Google-Attributes: gid103376,public From: peb@tcontec.demon.co.uk ("Paul E. Bennett") Subject: Re: Pratt & Whitney's Embedded Software - CMM Level 3! Date: 1999/04/20 Message-ID: <924570489snz@tcontec.demon.co.uk> X-Deja-AN: 468646878 References: <371B6EC8.36B9C247@pwfl.com> <7fftel$6po@drn.newsguy.com> <371B9A5E.2804AC27@pwfl.com> X-Complaints-To: abuse@demon.net X-Mail2News-Path: news.demon.net!tcontec.demon.co.uk X-Trace: mail2news.demon.co.uk 924599171 mail2news:28156 mail2news mail2news.demon.co.uk Organization: Transport Control Technology Ltd. Reply-To: peb@tcontec.demon.co.uk Newsgroups: comp.lang.ada,comp.realtime,comp.arch.embedded Date: 1999-04-20T00:00:00+00:00 List-Id: In article <371B9A5E.2804AC27@pwfl.com> diespammer@pwfl.com "Marin David Condic" writes: > ..................... Surrounding this idea, we got > ambitious and decided that all the diagrams should be assembled in > documents to meet Mil-Std-2167a requirements - hence a full-up editing > system was born. Beyond that, all documents (requirements, design, etc.) > needed to be pulled together into a "system" so that we could > automatically build the code, build the document set, etc. and preserve > all the artifacts. Up until this point you haven't mentioned what model you are using except for the reference to a standard that not all of us will be familiar with. Even if we all knew what was in Mil-Std-2167a, this has said nothing about how you met those requirements. yet, you seem to be in the dragging together of project information for coding stage already. > From there, a system needed to fall under configuration management and > change control, so the concept expanded even further. A system could be > baselined and a new one started from there. Any user could submit a > change request against a system and the whole flow of the CR was > automated - right to the point of being able to identify every artifact > that was impacted by the CR and every CR that went into changing a > particular version of a system or document. I find it incredible that the first form that many reach for is a Design Change Request. I have no such form in my process. I use the following in (more or less) this order. * Problem Report Form, to record the fact that aproblem exists. * Design Change Proposal, to establish what may be done about the problem. * Work Instructions, to ensure the correct implementation of a change. I also have a Review Report Form that acts as a coversheet for the issues raised as the result of a technical review meeting (of which there are four in my process loop). Incidently, getting it right first time will only visit two reviews, the other two are in the change path. > What it all meant was that we had laid the groundwork in software to > support an entire process from start to finish which was a) written > down, b) repeatable, and c) provided monitoring of the process for > continuous improvement. We didn't know anything at the time about CMM > because - I think - it didn't quite exist yet. (We're talking about 1989 > when this whole thing got started). It was a fortuitous turn of events > that had the great minds at CMU thinking exactly like the great minds at > Pratt & Whitney. :-) When you say start to finish, I take it you have your contracts under CM as well. > After quite some time flogging everyone into adopting our process and > gradually improving and increasing the automation, folks began to come > around and see that we really had something here. We've got metrics > dating back to the start of the project which have shown consistent > improvement of productivity and reduction of defects. The whole effort > has been a long and gradual improvement, but in the end, it got us the > Level 3 certification and, more importantly, a safer, more reliable > product at a lower cost. If you were to cost out the effort of development deployed in automating the aspects of the system you have dealt with, what is the per seat cost? This will be an important indicator for many. Commercial packages for a RCS and CM system that I have seen range as follows (per seat costs):- Cheapest - UKP 230 - not worth looking at really as a lot of work was required to configure for use. Most Expensive - UKP6000 - very complex and requiring huge amounts of vendor support. Also very expensive to establish the exact sales configuration. Best Value - UKP1000 - Up and running in a day for two users but rest of team within the week (including the initial training time). Required some time entering and adjusting roles data for the team. Rarely is a home grown effort at RCS and CM tools cost effective. However, as you seem to have been on the early wave I can understand your company doing it themselves. So how do you compare to the available commercial packages now. -- Paul E. Bennett ................... Forth based HIDECS Consultancy Tel: +44 (0)7971-620145 Going Forth Safely