Software bloat (Larkin was right)
in [Design]
|
Prev: OrCad/ question
Next: Capture hierarchy
From: John Larkin on 24 May 2010 18:18 On Sat, 22 May 2010 22:54:01 -0700, D Yuniskis <not.going.to.be(a)seen.com> wrote: >Hi John, > >John Larkin wrote: >>>>> There's nothing that says part numbers have to be positioned >>>>> on shelves in sequential order. You track the *location* >>>>> of a part as part of your inventory control system. So >>>>> that you are free to put things wherever is most convenient. >>>> Cool. Every part gets a location number that determines their location >>>> and order on the shelves. So why isn't that the part number? >>> Why *should* it be? Do you seat your children around the >>> table by age? alphabetically? >> >> If I had 4900 children, I probably would. > >And your cars are parked in the driveway sorted by date >purchased? Or, alphabetical by make/model? Or, VIN number? If I had 4900 cars, I would. But if I had 4900 cars, I could sell 4899 of them and retire. > >>>>> Let the machine figure out where something is. Let it >>>>> print your pick list in an order that *it* knows is >>>>> efficient (because *it* knows the physical layout of >>>>> your stockroom and can order the part numbers so you >>>>> don't wander around looking for numbers "in numerical order"; >>>>> sort of like preparing a grocery list -- you list the eggs >>>>> with the butter so you can pick up both of them in "Dairy") >>>> So the pick list is in sequential order by location, and the parts are >>>> ordered on the shelves by location number. Great idea. >>> Do you *really* think DigiKey has all of their parts on >>> shelves in numerical order? > >Well, we all like to marvel at how *wonderful* DigiKey's >search capabilities are -- even those exposed to the public. > >So, DO YOU THINK THEY HAVE THEIR PARTS ON SHELVES IN NUMERICAL >ORDER? I bet they cluster similar parts by part number. I've been in distributors' stock rooms, and they did. > >>>> So what use is the part number now? >>> Because you don't put "room 4, shelf unit 13, bin 5" on >>> a bill of materials! >> >> No, I put "231-2170" for a 1N5247B. And I can go to the stockroom and >> walk right up to the bin, knowing that number. It's the stock number >> *and* the location. Isn't that just too clever? > >Do your engineers think and talk about "using a 231-2170" >in a circuit? Or, do they think and talk about using a >"1N5247B"? (if the former, I'm sure that's a great skill >that will server them well at *other* companies!) They usually say "17 volt zener." We need our own stock number so that it has a known format, and so that we can specify multiple manufacturers and multiple MPNs to satisfy our requirements. > >You've done what most small companies do: one guy wants >to keep track of things IN HIS HEAD so he creates a numbering >system that lets him (try to) do so. As more and more >parts fall under that system, exceptions start to creep >in ("We'll just store the big stuff over in the next aisle... >we'll all *know* that because it's BIG STUFF"). He ignores >the '9xxx' syndrome because he's got too much invested in >the system already and minimalizes the issues that are >consequential to that choice. No one guy designed our system; we did it together. Everybody loves it. The only complaint we've had is that it would be better to minimize the occurrence of adjacent numbers that are nearly alike, because that can lead to picking errors. I wanted to add a checksum sort of letter, or maybe a random color code, to each part for pick redundancy. > >As new technologies emerge, he *bends* the system to try >to accommodate them -- there's always an "out" that can >be rationalized into the system. Time is spent coming >up with workarounds, standards, etc instead of being spent >on things that actually contribute to the *real* business. It's holding up fine so far. We left lots of hooks for expansion. > >>> You put *things* in *places*. Parts in locations. >>> You associate a part with a location in much the >>> same way that you associate a part with a subassembly. >>> >>> Part numbers are just ways to identify parts. >>> >>> Location numbers are just ways to identify locations. >>> >>> If you reorganize your stock room, should all of the part >>> numbers change? But, clearly the *locations* of all >>> of those parts must change! So, your desire to arrange >>> parts contiguously by part number forces your stock room to >>> have a particular shape. >> >> No. It forces groups of parts to be locally ordered. What's wrong with >> that? > >Why *should* the part number be the criteria for defining >placement of a part? The part number is derived according to a rules document. It just happens to put resistors and capacitors into order by value. It also does things like put all SMA connectors together. You don't approve of that? You'd rather see all the parts mixed at random so you can demonstrate the power of computer databases to look things up? We've had *affordable* computers for >nearly 40 years now. A machine can keep track of *two* >numbers just as easily as *one* (since it already has to >keep track of *vendor* part numbers as well!). I'm not interested in how hard computers need to work to find parts; I'm interested in how easy it is for people to use. People have been more expensive than computers for some years now. > >You're dealing with a false economy. I'm manufacturing electronics. You're a programmer, right? > >I had an employer who wanted everyone to program Z80's >using "split octal" (e.g., 0xFFFF is 377377). His rationale >was that you could easily decode -- or ENCODE -- the >instruction IN YOUR HEAD when you saw the opcode expressed >in this form. The PDP-11 instruction set was designed for octal. A double-operand 16-bit instruction was, in octal, BCSSDD where B=byte flag, C is opcode, SS is source register+mode, DD is destination reg+mode. Thing of beauty. Concatenating two bytes into one word was admittedly awkward. > >"Um, don't we have symbolic assemblers and debuggers to do >that thinking *for* us?? So we can concentrate on solving >engineering problems and let the machines handle the >trivial tasks??" > >>> (Or, do you have a little sign located between part #0999 >>> and #1257 saying "1000-1256 located down the hall"?) >> >> One of our part attributes, in the database, is "location". S means >> stockroom. B1 means one area of the basement, where we have stuff like >> chassis and power transformers. I think we have 4 or 5 named areas. > >So, why didn't you put those designators *in* the part >number? Because things can move around. As I've pointed out, the part number is based on the part functionality and value. We just stash them physically by part number. I.e., you have already acknowledged that there >are "exceptions" to your nice "ordered by part number" >scheme. Here come the '9xxx's... :> Sure. If we get a deal on a box of 20,000 pulse transformers, they won't fit on a shelf in a bin. So we stash some or all of them somewhere else. Being dogmatic is for dogs. > >> But everybody pretty much already knows that big transformers are >> downstairs; if you don't, look it up. The parts in the stockroom, the >> electronic stuff, wraps around the room in part number order. Hardware >> is in a carousel sort of thing on the production floor. A lot of >> people, production and engineering, designed the system. It works >> really well, much better than anything I've ever seen elsewhere... and >> I've seen some horrors. > >When I walk into a library, the *first* thing I do is go to >the card catalog. *It* tells me where the item that I am >interested in is located. Then, I look at the signs on >the stacks to figure out which stack will contain it. I don't. I browse. Besides, most "card catalogs" have been replaced by incredibly obtuse software running on seldom-available, often-broken PCs. Card catalogs had impressive multi-port parallel access features. > >The items aren't stored alphabetically. By title or author. >They aren't even stored in the Dewey Decimal system anymore! >I can roam around the library and try to get a feel for >where things are. But, this takes time. (most employers >have to pay people for their time :> ) The shelves have these cool metatags like GARDENING and MYSTERIES. > >OTOH, if "something" tells me an item is in a particular >location, then all I need is a map of where those locations >are to find the item. > >> If I need an SMB connector, I look up any one and then go browse the >> shelf where all the other SMBs are located. They are all together, >> because SMB connectors have a defined part number prefix. Resistors >> and capacitors and inductors are together, ordered by value. That's >> nice too. > >And if *I* need an SMB connector, I issue a query: > CONNECTOR and SMB Our query looks like CON++SMB. We didn't use "and" because there's a company called AND. >not only will this tell me which connectors have had >part numbers assigned to them, but it will also tell me >quantities (instead of having to count them "by eye"), >whether any are "held" for an upcoming build, when the >last buy was, how much they *cost* (since I can't look >at two connectors and decide which is most economical >based solely on appearance and "feel"), etc. > >I'll get more information *and* know where they are. Sure. We do that. Also where-used, and datasheets, and engineering notes, pictures, whatever. > >>> I'm deploying exactly this sort of system currently. >> >> Well, good luck, whenever you get it done. > >The numbering system has been working for almost 25 years. >As I said elsewhere, I learned this lesson before I got >out of college :> I've since seen many companies go >the "let me keep it in my head" approach and struggle >when The Guy retired, or couldn't anticipate a new >teechnology ("Gee, what order should I put these >CPus in? Numerical by vendor's part number? Alphabetical >by vendor name? Should I put all the peripherals for >a particular processor with that processor? Or, should >I impose some sort order on the peripherals -- UARTs >together, Timers together, etc.? Oh, my! I never >dreamed they'd make ethernet controllers in chips! What >category do I put those in? I'm sure the folks who >come browsing through the stockroom to see what sorts >of chips we have BEFORE they design their processor >boards will want all of them in *some* order...") > >The mistake *I* made when I did this was failing to put >*everything* into the system. Treat *everything* as >a number (contracts, screws, semiconductors, source >code, binaries, clients, informal documents, etc.) instead >of just the obvious things (screws, code, etc.). If >I wanted to, I can add them now -- since I don't have to >go back and "make room" in the numbering system (just >grab the next N part numbers! :> ) It's just not worth >the effort. Instead, I'll have to keep several different >"tracking systems" (contracts, "parts", contacts, etc.) >instead of being able to unify *all* of it. > >I am now porting it to a different hardware and software >platform and "adding automation". None of that changes >the core logic in the system -- just the way you get >data in and out. (e.g,, so a forklift operator can >access the data without getting off the truck) > >>> Everything has an identifier. EVEN PEOPLE! >> >> Saves having to remember all those pesky unstructured "names" things. >> "Good morning 34719622. Can I get you a cup of 7192453?" > >How do you know *who* pulled the parts for a build? It's on a production traveler. Not that it matters. >Or, designed a circuit? Name's on the drawing. Do you refer to designers by number? > Or, who the customer is? We have lots of customers for most products. Or >the vendor? Or, do you not really track those sorts >of things? For each part we buy, and that has a Highland stock number, we have a list of acceptable manufacturers and their part numbers. Duh. > >>> The RDBMS puts the smarts to the system. Map an identifier >>> into an "object type": "What is 12345678?" "12345678 is a person" That's sort of repulsive. Or autistic maybe. >>> >>> Based on that object type, figure out what the actual object's >>> characteristics are. "Please describe person 12345678" >>> "12345678 has brown eyes, is 6' tall, male, etc." >>> >>> When I need an identifier "anywhere", I impose criteria >>> on what sort of object I expect in that "application". >>> So, for a timeclock application, I expect the identifier to >>> map to "PERSON" and not "RESISTOR". >>> >>> Using this sort of approach, I can have many different classes >>> of objects and allow the descriptions for those objects to >>> be intelligently *parameterized*. >>> >>> E.g., create a new part: 55667788 >>> What is it: (multiple choice) RESISTOR >>> OK, since it is a resistor... >>> What value: 4K7 >>> Tolerance: 5% >>> Rating: 1/4W >>> etc. >>> >>> Create a new part: 55667789 >>> What is it: SCREW >>> OK, since it is a screw... >>> Head type: clutch >>> Thread pitch: 32 >>> Size: 6 >>> Length: 1" >>> >>> Create a new part: 55667790 >>> What is it: PCB >>> OK, since it is a circuit board... >>> Schematic: 12342345 (which must be of type "SCHEMATIC") >>> Artwork: 45567788 (which must be of type "ARTWORK") >>> BOM: 45324323 (which must be of type "BoM") >> >> Yikes. Now you need a product structure document (or database) to tell >> you which schematic goes with which bare board and which assembled >> board. And where they are all kept. And what drawings (of type >> "drawing") and parts lists (of type "parts list") apply. >> >> Been there, done that, never again. > >How do you deal with a vendor calling to inquire about a >system you built for him 5 years earlier which, perhaps, >has developed a bug. Well, first I have to type on this "computer" thing to look up the manual, to figure out what he's probably doing wrong. If he knows his unit's serial number, I can determine what code he's running and look up its status. Or, perhaps the customer wants an >*identical* system (because the last one has already >been through FDA validation -- "functionally equivalent" >is not the same as "identical"). We do that all the time. We use the same sorts of systems all the aerospace and mil folks use. That's my background. > >You need to know all of these things in order to be able to >reproduce *the* system you sold him. > >You also need to know exactly which version of every >software function/routine you used. Which of each tool >used to create it. Which configuration of each OS, etc. Sort of. We don't ship OSs, just hard embedded things. > >What do you do when a customer calls and points out a >*bug* in your product? If you can narrow it down to >a routine, how do you know which *other* products may >be affected? How do you know which other builds >contained *that* "software component" (not just one >that had that "function name")? Or, do you just >wait for other customers of other products to call >with problems later? It's not all that difficult. > >I design boards that typically serve multiple purposes. >Lots of stuffing options. Sometimes those options are >exposed to customers (if they are going to support >the product "in the field"). Sometimes, they are >just unnecessary details ("No, you don't need to >know that cutting this foil switch will allow you >to use a larger memory device in place of two smaller >ones"). Cutting foil switches? In 2010? Are you serious? > >The same sort of "I-want-to-keep-it-all-in-my-head" >mindset leads to equally silly numbering systems: > SCH-12345 > BOM-12345 > PCB-12345 >Wow! Isn't this great?! I can come up with a part >number for the schematic for this board without having >to turn on the computer! Gee, welcome to the 21st >century!! Sch 22S450B PCB 22D450B Assy 22A450-12B Code 22E450E FPGA 22V450C Cool! > >You then have to deal with "variants" -- fabricate NEW >part numbers (schematic, PCB, etc.) for each of these >variants. "Gee, lets' put a dash after the part number!" That's the way they build airplanes. They use even dash numbers as mirror images of odd ones. You know, left wings and right wings. >(Of course, this doesn't help you figure out the significance >of each particular "dash number" -- "Oh, we can always >look that up -- on the computer!") It's in the price list. > >Instead, I create *really* new part numbers whenever I need >one. They are inexpensive. I can apply a single "board" with >different silkscreens to different finished assemblies -- and >give each a separate schematic *or* have them share a common >schematic. The part number for the BOM need not be related >to the SCH *or* the PCB *or* the final assembly. > >When you purchase a "final assembly", you are purchasing a >tangible piece of hardware. The schematic, artwork, BOM, >etc. are all *different* entities. One doesn't imply the >other. (i.e. a schematic, BOM, piece of FR4, etc. can >exist in the absence of any of the others!) Sure. That's necessary and obvious. > >If I have to "sell" a documentation package, then I create >a P/N for that package -- which could just be a bunch of >paper (or files). It doesn't have to include the actual >physical "device". Ditto. > >> I bet that if you have a shell-and-pin connector, the shells are kept >> far away from the pins because "shell" and "pin" are different types. > >The shell is kept wherever the hell you want REGARDLESS OF >IT'S PART NUMBER! *Your* scheme FORCES the shell and the >pin to have adjacent part numbers because you *want* them >to be adjacent and can only do so *if* they have consecutive >(or nearly so) part numbers. *I* can put them in any order >I want -- pins *before* or after the shells AS I SEE FIT >(regardless of their part numbers) -- because I dissociate >the part number from its location. Sounds like it's a wonderful intellectual exercise for you and a PITA for the production people. John
From: D Yuniskis on 24 May 2010 21:44 Hi Michael, Michael A. Terrell wrote: > D Yuniskis wrote: >> Hi Michael, >> >> Michael A. Terrell wrote: >>> D Yuniskis wrote: >>>> Most of my old paperbacks have completely bogus "house numbers" >>>> on them. Of course, there were *lots* of publishers around >>>> at that time (since gone belly-up or "absorbed") so you didn't >>>> need very many digits in *your* "book number". And, no effort >>>> to make the numbers share a universal, common "number space". >>> I'm in the process of inventorying my book collection. Less than half >>> have an ISBN. >> Not surprising. I suspect at least half of my books (in terms of >> quantity) have none. >> >>> Some that do have the ten digit, while the newest have the >>> newer 14 digit version. Then there are newer books that were sold by >>> book clubs, with no ISBN or publisher identification. :( >> *No* publisher? Not even written in text?? E.g., all of my >> (kangaroo) "Pocket Books" have bogus numbers but still declare >> a publisher. > > Sorry, I meant no publisher's stock number, like older books had. Ah, OK. > They just named the publisher and listed it as 'Book Club Edition' which > makes it hard to find any additional information on some books. I still Understood. This is particularly true of "short run" titles. > have another 1000 books to inventory. :( Some are stacked to the > ceiling in Banana boxes. Every shelf in my library is full, and I am > considering turning the bedroom next to it into another, larger library. Yikes! I pruned all of my databooks and paperbacks some years ago. Cherry picked what I wanted to hold onto and got rid of everything else (donated the paperbacks to local library -- they wouldn't be interested in databooks! :> ). I tend to only hold onto the books (paperbacks) that I regularly re-read. Otherwise, I'd rather use the space for something else ;-) Of course, got rid of a few things that I really wish I hadn't. But, you don't figure that out until *after*! :< (Really wish I had kept the "EYES ONLY" specs for the Raster Graphics Processor -- though I don't think they ever made it but it had some interesting ideas). Sure would be nice if there was an "un-discard" button I could click! :-/ The books that I accumulate now get "recycled" at the end of the year. I've found a few folks who like certain parts of my "library" so I set things aside as appropriate. The rest go to the library. (I read about 25,000 pp annually "for entertainment") Lately trying to get the local library to buy or ILL the books that I want to read as it gets pricey to buy them, read them once and then "discard" them. :< The downside of donating paperbacks is they have an incredibly short lifespan once "in circulation". It's like a death sentence! :<
From: Michael A. Terrell on 24 May 2010 23:42 D Yuniskis wrote: > > Hi Michael, > > Michael A. Terrell wrote: > > D Yuniskis wrote: > >> Hi Michael, > >> > >> Michael A. Terrell wrote: > >>> D Yuniskis wrote: > >>>> Most of my old paperbacks have completely bogus "house numbers" > >>>> on them. Of course, there were *lots* of publishers around > >>>> at that time (since gone belly-up or "absorbed") so you didn't > >>>> need very many digits in *your* "book number". And, no effort > >>>> to make the numbers share a universal, common "number space". > >>> I'm in the process of inventorying my book collection. Less than half > >>> have an ISBN. > >> Not surprising. I suspect at least half of my books (in terms of > >> quantity) have none. > >> > >>> Some that do have the ten digit, while the newest have the > >>> newer 14 digit version. Then there are newer books that were sold by > >>> book clubs, with no ISBN or publisher identification. :( > >> *No* publisher? Not even written in text?? E.g., all of my > >> (kangaroo) "Pocket Books" have bogus numbers but still declare > >> a publisher. > > > > Sorry, I meant no publisher's stock number, like older books had. > > Ah, OK. > > > They just named the publisher and listed it as 'Book Club Edition' which > > makes it hard to find any additional information on some books. I still > > Understood. This is particularly true of "short run" > titles. Some of the book club editions are the standard volume, printed on newsprint grade paper so the printing costs are lower. They aren't intended to have a long life. Unfortunately, some collections of short Sci-Fi stories can only be found in that format. > > have another 1000 books to inventory. :( Some are stacked to the > > ceiling in Banana boxes. Every shelf in my library is full, and I am > > considering turning the bedroom next to it into another, larger library. > > Yikes! I pruned all of my databooks and paperbacks some years > ago. Cherry picked what I wanted to hold onto and got rid of > everything else (donated the paperbacks to local library -- they > wouldn't be interested in databooks! :> ). I tend to only hold > onto the books (paperbacks) that I regularly re-read. Otherwise, > I'd rather use the space for something else ;-) > > Of course, got rid of a few things that I really wish I hadn't. > But, you don't figure that out until *after*! :< (Really > wish I had kept the "EYES ONLY" specs for the Raster Graphics > Processor -- though I don't think they ever made it but it > had some interesting ideas). Sure would be nice if there was > an "un-discard" button I could click! :-/ > > The books that I accumulate now get "recycled" at the end of > the year. I've found a few folks who like certain parts of > my "library" so I set things aside as appropriate. The rest > go to the library. (I read about 25,000 pp annually "for > entertainment") Lately trying to get the local library to > buy or ILL the books that I want to read as it gets pricey > to buy them, read them once and then "discard" them. :< > > The downside of donating paperbacks is they have an incredibly > short lifespan once "in circulation". It's like a death > sentence! :< Does your area have a 'Friends Of The *** Library' group? They accept & resell used books to help the library they support. That allows the library to spend the money on new books or other items they need for their members. Here is an example: http://www.friends-of-belleview-library.org/ -- Anyone wanting to run for any political office in the US should have to have a DD214, and a honorable discharge.
From: D Yuniskis on 25 May 2010 08:57 Hi Michael, Michael A. Terrell wrote: >>> Sorry, I meant no publisher's stock number, like older books had. >> Ah, OK. >> >>> They just named the publisher and listed it as 'Book Club Edition' which >>> makes it hard to find any additional information on some books. I still >> Understood. This is particularly true of "short run" >> titles. > > Some of the book club editions are the standard volume, printed on > newsprint grade paper so the printing costs are lower. They aren't > intended to have a long life. Unfortunately, some collections of short > Sci-Fi stories can only be found in that format. Understood. I have a copy of _Farewell to the Master_ (ties with _The Sentinel_ as my favorite short story) in an October 1940 issue of _Astounding Science Fiction_. I rarely open it for fear of the pages "breaking" (I have another copy in an anthology that is more resilient). [If you enjoy SF, you might enjoy _The Lost Worlds of 2001_ as an insight into the "film" (read it and you'll see why I say "film" instead of "book" :> amusing bits of history) ] >>> have another 1000 books to inventory. :( Some are stacked to the >>> ceiling in Banana boxes. Every shelf in my library is full, and I am >>> considering turning the bedroom next to it into another, larger library. >> Yikes! I pruned all of my databooks and paperbacks some years >> ago. Cherry picked what I wanted to hold onto and got rid of >> everything else (donated the paperbacks to local library -- they >> wouldn't be interested in databooks! :> ). I tend to only hold >> onto the books (paperbacks) that I regularly re-read. Otherwise, >> I'd rather use the space for something else ;-) >> >> Of course, got rid of a few things that I really wish I hadn't. >> But, you don't figure that out until *after*! :< (Really >> wish I had kept the "EYES ONLY" specs for the Raster Graphics >> Processor -- though I don't think they ever made it but it >> had some interesting ideas). Sure would be nice if there was >> an "un-discard" button I could click! :-/ >> >> The books that I accumulate now get "recycled" at the end of >> the year. I've found a few folks who like certain parts of >> my "library" so I set things aside as appropriate. The rest >> go to the library. (I read about 25,000 pp annually "for >> entertainment") Lately trying to get the local library to >> buy or ILL the books that I want to read as it gets pricey >> to buy them, read them once and then "discard" them. :< >> >> The downside of donating paperbacks is they have an incredibly >> short lifespan once "in circulation". It's like a death >> sentence! :< > > Does your area have a 'Friends Of The *** Library' group? They Yes, I laid out the last newsletter for ours. I'm currently chasing down a (free) source of "removable" self-adhesive labels for them to use to mark prices on the books they sell on their "used book cart". > accept & resell used books to help the library they support. That > allows the library to spend the money on new books or other items they > need for their members. Here is an example: > > http://www.friends-of-belleview-library.org/ Correct. Though our group will *not* buy books with proceeds from donations. The thinking is that the books *circulate* so don't benefit *this* branch library as much as, for example, aprons for the staff, a new table for the conference room, etc. <shrug> Very few donated books end up making their way into circulation because of their "flimsiness". Unfortunately (?), most of my "hard bound" texts are keepers :>
From: Spehro Pefhany on 25 May 2010 12:03
On Tue, 25 May 2010 07:53:18 -0700, D Yuniskis <not.going.to.be(a)seen.com> wrote: >Hi John, > >John Larkin wrote: >>>> If I had 4900 children, I probably would. >>> And your cars are parked in the driveway sorted by date >>> purchased? Or, alphabetical by make/model? Or, VIN number? >> >> If I had 4900 cars, I would. But if I had 4900 cars, I could sell 4899 >> of them and retire. > >I'd put the car I drive most closest to the *door*! >I wouldn't get hung up "sorting" them based on some >dubious and changeable criteria. It would suck to buy a new Alfa Romeo and then have to spend weeks moving 4800 cars in order to maintain alphabetical order. |