|
|
|
Thread Tools | Display Modes |
#11
|
|||
|
|||
Software "hardness". was: Signal-to-noise in sci.space.*
Doug... said:
Yep, there are good reasons for some of that "bureaucratic bs." One of the genuine challenges of engineering management is to impose the correct bureaucratic process. Just adding a new layer of paperwork every time a mistake isn't caught can quickly stifle progress. And in my experience, programmers and engineers who insist that documentation requirements are unecessary because *they* already know the details of their own systems are usually either a) lazy or b) trying to maintain the ego-boosting position of being the only person who understands something no one else understands. Both situations often result in disaster. I'll add c) they lack the engineering maturity to realize the importance of the documentation. I've seen a project die or nearly so when one or two key people left the team. The engineering talent is usually replaceable, but the in-depth knowledge of how the software functions was limited to those one or two minds. The current flavor of the month in commercial software development is eXtreme Programming. XP is a counter-reaction to too much process and paperwork. The focus is on building a small, lightweight, agile team that can respond to changes quickly and effectively. Because they respond so well to changes (the XP proponents claim), there is no need for excessive written documentation, nor time consuming reviews of completed programs. The XP philosophy is "the tests all run without error, so the software must be right. I'm watching XP with fascination and dread. It is a tool that has some value in some circumstances, but XP is being used on tasks that are well beyond XP's capabilities. -- Kevin Willoughby lid We'd spend the remaining time trying to fix the engine. -- Neil Armstrong |
#12
|
|||
|
|||
Signal-to-noise in sci.space.* (was Any legal basis to take Maxson down?)
On Sat, 16 Aug 2003, Jorge R. Frank wrote:
There is no evidence to support this view. I am talking about redirecting threads from s.s.shuttle - and *only* s.s.shuttle - to s.s.moderated. The moderators prune the noise, leaving only the signal. The end result is that the traffic on s.s.moderated will equal the sum of 1) its current traffic, plus 2) the traffic on s.s.shuttle, minus 3) the noise on s.s.shuttle, plus 4) the signal from any newbies that were allegedly previously intimidated from posting on s.s.shuttle. In my opinion, there is *no way in hell* that 4) is greater than 3), therefore the total traffic *will* go down. Undoubtedly the signal-to-noise will go up. *Way* up. On the same topic... I would suggest that people be vocal and clear about redirecting the threads that are of interest to s.s.m. That will highlight to the newbies what is going on, rather than have them get confused by seemingly random changes in the newsgroups line. -- Leach |
#13
|
|||
|
|||
Signal-to-noise in sci.space.* (was Any legal basis to take Maxson down?)
Leach wrote in
: On Sat, 16 Aug 2003, Jorge R. Frank wrote: There is no evidence to support this view. I am talking about redirecting threads from s.s.shuttle - and *only* s.s.shuttle - to s.s.moderated. The moderators prune the noise, leaving only the signal. The end result is that the traffic on s.s.moderated will equal the sum of 1) its current traffic, plus 2) the traffic on s.s.shuttle, minus 3) the noise on s.s.shuttle, plus 4) the signal from any newbies that were allegedly previously intimidated from posting on s.s.shuttle. In my opinion, there is *no way in hell* that 4) is greater than 3), therefore the total traffic *will* go down. Undoubtedly the signal-to-noise will go up. *Way* up. On the same topic... I would suggest that people be vocal and clear about redirecting the threads that are of interest to s.s.m. That will highlight to the newbies what is going on, rather than have them get confused by seemingly random changes in the newsgroups line. I agree completely. Along the same line, I don't plan on redirecting threads that are crossposted into sci.space.shuttle from people who are obviously posting from other groups, like the HST/JWST thread I replied to today. There is a good chance, in these cases, that the poster isn't subscribed to s.s.m. -- JRF Reply-to address spam-proofed - to reply by E-mail, check "Organization" (I am not assimilated) and think one step ahead of IBM. |
#14
|
|||
|
|||
Signal-to-noise in sci.space.* (was Any legal basis to take Maxson down?)
"Cameron Dorrough" wrote in message
... Thanks, Rich - that does clear a few things up. I've been lurking around s.s.t the most, due to my interest in things technical, and very often posts can take more than 24hrs to reach me. The issue isn't just *my* posts, but *any* posts to the NG. My ISP has been having a few hassles so that may be it - I think the timezone issue might be another one (the moderators being asleep sometimes when people over here post.) It's 9:40am here - I'll see how long this takes to get there.. ;-) It just appeared on my news server - it's 12.50pm local time. ..but it hasn't appeared on the Google Groups yet (work that out!). Hmmm. 3 hours 10 minutes and some seconds. Must be a record! :-( Cameron:-) |
#15
|
|||
|
|||
Software "hardness". was: Signal-to-noise in sci.space.*
Kevin Willoughby wrote:
James Summers said: In some ways, it was easier then. We had a lot less bureaucratic bs to put up with. In theory, the purpose of the bureaucracy is to prevent important details from slipping through the cracks. How did you keep with rate of change in requirements during Apollo? A bureaucracy kept book on all that. But keep in mind the two faces a bureaucracy. One that is nominally on your side in pursuit of a common goal is a good thing from the worm's eye view. One that insists on schedules, budgets, performance, etc... is not. D. -- The STS-107 Columbia Loss FAQ can be found at the following URLs: Text-Only Version: http://www.io.com/~o_m/columbia_loss_faq.html Enhanced HTML Version: http://www.io.com/~o_m/columbia_loss_faq_x.html Corrections, comments, and additions should be e-mailed to , as well as posted to sci.space.history and sci.space.shuttle for discussion. |
#16
|
|||
|
|||
Software "hardness". was: Signal-to-noise in sci.space.*
The current flavor of the month in commercial software development is
eXtreme Programming. XP is a counter-reaction to too much process and paperwork. As someone who has done commercial software development (with and without XP), I would say that XP has much more process than what is typical in commercial software development. There are lots of people for whom XP is their first exposure to unit tests, acceptance tests, release planning, and the like. Of course there are people for whom "XP" just means "throw out what little paperwork we have and go back to cowboy coding" but that's people finding what they want to find, not people actually reading and applying the XP books and websites. |
#17
|
|||
|
|||
Software "hardness". was: Signal-to-noise in sci.space.*
It's in my reading queue to finish going through the mishap report for
Apollo 13 and look at the root cause for this. There would have been a parts stress analysis done. Was there an error in it? Was it not done? Did it not get updated for a change? Or was part of it not done since it hooked up to GSE? It's been a while since I read the mishap report, but basically the voltage got changed from 28V to 65V (or whatever the exact numbers were) and not everything got updated accordingly. It didn't show up on the ground because it was a "sometimes fail" kind of deal (I believe). But if you do get around to reading the mishap report (which I believe is on-line, for anyone interested), please do offer a more detailed description of the various tests/analyses and where things got missed. The mishap report urged more attention to change control and my impression is that this is one of the sources of much of the paperwork that we see today (with the various good and bad consequences thereof). |
#18
|
|||
|
|||
Software "hardness". was: Signal-to-noise in sci.space.*
Jim Kingdon said:
The current flavor of the month in commercial software development is eXtreme Programming. XP is a counter-reaction to too much process and paperwork. As someone who has done commercial software development (with and without XP), Likewise. I would say that XP has much more process than what is typical in commercial software development. No. XP eschews the concepts of Product Requirements (telling stories just isn't the same), pretends you don't have to design something before building it (note the heavy emphasis on refactoring), assumes that it is better to read 50 pages of code than one page of prose (note the XP philosophy of minimum documentation), and misunderstands the distinction between Quality Assurance and testing (unit tests are necessary, but far from sufficient). If you read what Kent Beck has written about his creation of XP, it is clear that it was a counter-reaction to a process-excessive environment. He went from one extreme to the other. Neither extreme is desirable. ObSSM: On the day the CAIB released their report, I read the section of Secret of Apollo that did a compare&contrast on the Ranger and Mariner programs. Oversimplifying the author's views a bit, here we had two series of roughly similarly complex spacecraft, built and flown by the same organization (JPL), with very different results. Ranger was a disastrous failure -- six flights(*), each and every one a complete failure. Mariner was a success. The difference? Marriner had a comprehensive process, including serious change-control and configuration-management, with strict attention to be sure that "as built" exactly matched "as designed", and the design was based on complete specifications. In today's vocabulary, Mariner was process heavy and Ranger was XP. There are lots of people for whom XP is their first exposure to unit tests, acceptance tests, release planning, and the like. Sure, there are newbies in any field. Hey, I was one myself, long long ago. But the fact that it introduces new concepts to newbies isn't the same thing as saying that XP has more process than typical commercial environments. * There were follow up Rangers that were successful. Ironically, if it hadn't been for the success of Mariner, JPL might have been shut down before Ranger 7 flew. -- Kevin Willoughby oSpam Imagine that, a FROG ON-OFF switch, hardly the work for test pilots. -- Mike Collins |
#19
|
|||
|
|||
Software "hardness". was: Signal-to-noise in sci.space.*
rk said:
Jim Kingdon wrote: It's been a while since I read the mishap report, but basically the voltage got changed from 28V to 65V (or whatever the exact numbers were) and not everything got updated accordingly. From what I recall, those numbers are about right and it was for the non-flight interface on the tank; the flight interfaces stayed at 28V, which is a semi-standard spacecraft voltage. Many years ago, I built a commercial DataBase Management System. That forced me to learn a bit about this history of DBMS's. The first significant commercial DBMS was IBM's IMS. Somewhere or other, I read that IBM acquired IMS from North American, who created it to keep track of the huge number of changes that they knew would be serious issue in building the Apollo CSM. The system worked pretty well, since the only non-trivial configuration management problem in the CSM that I can recall is the Apollo 13 oxygen tank. The mishap report urged more attention to change control and my impression is that this is one of the sources of much of the paperwork that we see today (with the various good and bad consequences thereof). As I've noted elsewhere, I'm following the successes and failures of eXtreme Programming with amusement and interest. XP really tries to minimize paperwork to its absolute minimum. To his credit, XP creator Kent Beck seems to understand that this works for, and only for, relatively small projects. Once you get a project team that can't all have lunch around the same table, you *can't* rely on informal chats, you have to adopt a more formal approach. Apollo was a hugh project with many subtle inrelationships among its components, so it required a lot of paperwork. Actually that sort of thing preceeded Apollo. It's a major theme in _The Secret of Apollo_ which I'm now reading That major theme is one of the issues I have with the book. Configuration management *wasn't* invented to build rockets. Johnson makes an eloquent case that configuration management is required for systems as complex as an ICBM, much less Apollo. (I mentioned his Ranger / Mariner case-study elsewhere. I should add his Corporal / Sergeant case-study further emphasizes this need.) But the claim that it was invented for big rockets just doesn't match my understanding of the history of technology. As a kid, I had to listen to my father explain the differences between the P-51B and P-51D, so clearly North American was doing some kind of configuration management during WW-II. I suspect (but can't demonstrate) the concept goes back at least to the introduction of mass production / standardized parts. -- and this book [...] is rapidly becoming one of my best. Mine, too. As a kid, I wanted to be an astronaut when I grew up. I've read a lot of books written (at least nominally) by astronauts. As I became more interested in engineering, I became fascinated by the hardware of Gemini and Apollo, which is described in many books. As I tried (unsuccessfully) to be an engineering manager, I wondered about the management of Apollo. *That* topic is woefully underrepresented in the literature -- Murray & Cox is great for what it is, but doesn't describe the management processes as seen by a manager. Secrets is teaching me a lot. Last weekend, my girlfriend dropped by and noticed the books on my living room table. Of that dozen books, two are space-related. She looked at Secrets of Apollo and commented that it was an expensive book. She picked up Collin's Carrying the Fire, and noted it cost less than half as much. I'm only about 1/3 of the way through the book, but I've gotten more than my money's worth of insight from it. -- Kevin Willoughby oSpam Imagine that, a FROG ON-OFF switch, hardly the work for test pilots. -- Mike Collins |
|
Thread Tools | |
Display Modes | |
|
|