On Aug 12, 2009, at 2:43 PM, DJ Delorie wrote: > >> A rule-based thing sounds tricky. I think most would be happy with a >> simple list. > > Well, the rules would be in the plug-in, not geda. Geda says "what > values work here?" and gets a list. Gsch2pcb tries to fill in blanks > by requesting defaults, and complains if it can't get one. How all > that happens, is undefined. > > In my case, what I do is try to see how many parts in my database have > the right attributes. If the answer is zero, fail. If it's one, use > that one. If it's more than one, the perl script picks one based on > some heuristics I wrote. > > But for gschem, imagine if you could "ee" a symbol, select > "footprint", and have it be multiple choice? I mean, the database > (somehow) knows which footprint that device is available in... no > point letting you fill in TQFP44 for an ethernet jack. > > Now, suppose you had a capacitor. Lots of footprints for those. Now > pick 0.1uF, there are fewer footprint options. Choose Kemet, you're > down to three options. Choose 10v (of the four voltage options you > now have ;) and you're down to one. Don't like those footprints? > Erase some of the other attributes, select the footprint you want, and > see what options you have for the erased ones. Hmmm, maybe you need > to choose a 16v instead. > > I get this idea from the digikey web page - the way they let you drill > down through part attributes until you get a small enough list to pick > one of them.
The Digikey web page has a lot of resources behind it. I use it a lot. It's the best in the business. But even with all that, it requires a lot of hand holding by the user. There are errors in the database (better retrieve the datasheet and check!). Some categories are hard to understand, some requirements not reflected. It's a time saver, but it's not anywhere close to full automation. It takes a lot of human time and work. Now, imagine it was integrated into gschem. That would be slick, but would it improve productivity much? I think not. The actual time spent cutting/pasting/transcribing the information into the attributes is a small fraction of the the time this process takes. That's not where the bottleneck is. Anything the gEDA project could implement would be much less effective. We don't have Digikey's resources. I think the real question for further improvement to gEDA is "where are the bottlenecks?". Not the "it's not as slick as tool xxx" or "it doesn't work the way I want it to" complaints, but the real timewasters where there is no reasonably easy way to automate the flow. John Doty Noqsi Aerospace, Ltd. http://www.noqsi.com/ j...@noqsi.com _______________________________________________ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user