JM> SDRC - just another overpriced, overdone program that is
 JM> never a finished product.
  
 IU> Interesting - the 3D version or their 2D version. (They  hang 
 IU> out in the UK near where I used to live and we have  had some 
 IU> good demos.) It's a big commitment to buy one  3D package "over" 
 IU> another - evaluated a few.
  
   Didn't know there was a 2D version. And they are "pricey".
  
 IU>  If anything I'd have thought the "never finished" bit applied  
 IU> to all major programs including Acad but especially the their  
 IU> modeller rendition ;-)
  
   Yes, so I look for tools that are as simple as possible. Already 
 been burned by Computervision.
  
 IU>  The biggest (commercial) pressure came from ProE and it  wasn't 
 IU> nice to use at all.
  
 I thought ProE was a good demo, very powerful for making 
 pictures/design, but manufacturing is a completely different story.
  
 IU>  Now are you saying that I should consider Acad as a serious 3D  
 IU> modeller? I want something slick and polished with a nice  
 IU> interface that knows about surfaces, etc.
  
   Well, yes, I say you should consider it, depending on what your 
 requirements really are. As for the slick interface, well that may 
 be what you want, but maybe not what you need. I got my AutoCAD R12 
 for $200 or so as a bona fide student in a night class. The whole 
 college tuition and all was about $1400. That got me into AutoCAD, 
 some instruction, and a couple of college hours. I tried to do some 
 work with DesignCAD 2D but it was more trouble than it was worth, 
 and I paid the same price ($200) for DesignCAD. In some respects 
 AutoCAD R12 surpasses Computervision capability that cost probably 
 about $1,000,000 a decade or so ago.
  
 IU>  I'm fed up with 2D modelling and having to draw complex 
 IU> surfaces  and sections (particularly in side view) far better 
 IU> these views  are auto drawn as soon as the depth information is 
 IU> given.
  
   I don't follow the phrase "auto drawn". You may mean parametrics. 
 That is a nice design feature.
  
 IU>  Far better that a late design change doesn't mean a nearly  
 IU> total redraw of /all/ views.
  
  I have yet to see a convenient way to go from 3D to a 2D drawing. 
 Fortunately, I usually just import geometry and machine it, rather 
 than need to dimension it myself. But I have done some design work 
 and it takes considerable forethought and additional layering in 
 different views to do the right kind of modeling that appears as a 
 2D object in each view.
  
  Solids make super pictures, very quickly. Blending of fillets is 
 often a nightmare, and sometimes just not supported in certain 
 cases. Unless you need mass properties, the solids representation 
 may not be of much real help. But this depends on the kinds of 
 toolpaths that are required. If you just need some "canned" routines 
 for running around mold surfaces it may be just fine. But if you 
 need 5 sided machining where you flip the part around on the table, 
 the graphics concept of copying and flipping the part geometry to 
 the machine reference system is not copacetic. What you get is 
 duplicated geometry, and as you mentioned, a problem in the case of 
 design changes. I prefer to use a single part representation, upon 
 which the tool path can be verified as a single setup, and the 
 complete part emerges from a solid piece of stock, and it only takes 
 one file to contain the part program, not multiple tapes or files. 
 Then to accommodate the orientation of the toolpath to the machine 
 tool I put transformation statements into the source file to flip 
 the punch file (G codes) of that particular section of part program 
 into the machine tool's coordinate system, at the appropriate 
 position. (I'm getting long winded here.) 
  
   Sorry, I don't use the cutesy surfaces much at all, one exception 
 being to check out a toolpath in a solids verifier program apart 
 from AutoCAD. The trick to me is to get a variety of programs that 
 do the individual functions well, (design, drafting, toolpath 
 generation, feedrate/auxillary function calculation, toolpath 
 verification, etc.) but that work together. When I couldn't find a 
 program to perform a function, I wrote it myself. For design and 
 drafting, and with a good bit of customization and lisp routines R12 
 is not bad.
--- FLAME v1.1
---------------