Re: gEDA-user: geda cygwin package

[Joerg <joergsch@xxxxxxxxxxxxxxxxxxxxx>]

al davis wrote:
> On Thursday 14 May 2009, Joerg wrote:
>> Not bug-for-bug, but as close to the industry standard as
>> practical. If you let them use something that has nice
>> features they get used to those, and then in industry where
>> they don't have those it becomes a problem. You can do just
>> about anything with SPICE, even simulate mechanical devices.
>> But one has to learn who to "kludge and cajole" SPICE to do
>> that.
> 
> It's pretty close.  In the development snapshot, and in the next 
> official release due  this summer, the whole user interface is 
> done by plugins, which can be made any way you want.  Perhaps 
> someday it will be bug-for-bug, with more than one, and still 
> have extensions if you ask.
> 
> Significant parts of industry are moving to Verilog-A.  None of 
> the cheap simulators have it, but Gnucap does.  Gnucap also  
> accepts Spectre netlists, a subset.
> 

I can only speak for analog. There, I can currently see nobody moving 
away from SPICE, and except for chip design most are migrating towards 
LTSpice for obvious reasons ($xxxx versus $0). So when they hire new 
engineers they prefer them to be familiar with LTSpice.


> As to those nice features ..  all tools have advantages and 
> disadvantages.  There are features you will find in one but not 
> another.  Students need to learn this.  So, having nice features 
> like Gnucap's interactive operation, extended probes, and the 
> ability to easily change circuits interactively or with scripts 
> is good.  The ability to directly enter a netlist without file 
> baggage is a big help at the beginning.
> 

Ahm, I used to enter everything by netlist and some of the old stuff I 
have used on LTSpice. It can do that, you'd be free to write a netlist 
there.


> On the other hand, the tightly integrated graphics of LTspice 
> and other PC simulators is in that category where they don't 
> really add to functionality, but become a crutch.  Then they 
> have a problem when the GUI isn't available, or more importantly 
> they have a task that is complicated enough that the GUI gets in 
> the way.  Those features are the ones to avoid.
> 
> So, it's LTspice that has "nice" features that they would be 
> better off without.
> 
> I start them with a netlist, then later they learn how to use 
> schematic capture as a way to generate a netlist.
> 

LTSpice can accomodate that. It works off of a plain old ASCII file. Any 
time I needed it to do something in a more traditional way like in the 
DOS days, it complied.


>> Even a (very) seasoned engineer had to ask about that three
>> days ago, how to set abstol and stuff and make it travel with
>> the file. Nothing wrong with asking.
> 
> Students need to learn that simulators don't always work.  In a 
> senior level course on analog design, it is reasonable to expect 
> that they will see a convergence failure, and need to mess with 
> abstol and stuff.  It might even be desirable for a simulator to 
> have a hidden mode that makes convergence worse to make sure 
> this happens.
> 

Yes :-)

Just like the flight simulator where suddenly the manifold pressure 
changes unpredictably right after take-off.


>>> Students need to learn to be flexible, and they need to
>>> learn to use computers effectively, not just by kicking the
>>> GUI a few times.  EE's, even analog designers, need to
>>> learn some serious programming.
>> They need to and they do, to some extent. They do not have to
>> become programming experts, else I might as well demand that
>> all CS guys fully understand Maxwell's equations because we
>> have to ;-)
> 
> Even the CS guys are not programming experts.  The EE's should 
> be able to work with unix, with the command line.  They should 
> be able to write programs to solve engineering problems.  They 
> should be able to administrate their own systems and write 
> scripts to solve their own problems.  They should be able to 
> install a program from source, and do some simple porting.
> 

Ok, here we differ. There comes a point where the amount of learning is 
plain impossible to cram into any brain in the given number of 
semesters. While, for example, it would be nice if every person holding 
a commercial driver license is able to design their own engine control 
unit this is not going to happen.

Did I ever write my own programs? Yes. Would my career have come to a 
screeching halt if I wouldn't have? No. I would have just paid someone 
to do it for me.


> At both universities where I was on the faculty, we got constant 
> comments from employers that the students need to learn more 
> programming.  We got those comments from accreditation reviewers 
> too.  We were not worse than average.  It's a widespread 
> problem.
> 

If by accreditation you mean ABET I better not comment, cuz it'll get 
ugly ;-)


>>> Too many schools don't do this.  In the extreme case, EE
>>> could become a dumping ground for students who can't make
>>> it in CS. Is that what you want?
>> That has IME never been the case, and won't be. None of the
>> EEs I know started out CS. And don't believe EE is easy, our
>> university had an EE flunk-out rate of around 75% plus. ME
>> and EE were the toughest paths there, some of the grueling 4h
>> written exams could make grown men shake in their boots.
> 
> Typically, the first year curriculum is pretty much the same for 
> all science and engineering.  The students usually have not 
> really have made up their mind yet.  They made a guess, but 
> that's all.  They move around depending on how the first year 
> goes.  It amazes me how many have said "I chose EE because I had 
> trouble with the programming course and want something else".  
> Then they learn too late that EE is hard too.
> 
> At a lot of schools, the EE program (and probably others) keeps 
> getting easier.  Many older EE professors are having a hard time 
> dealing with this.
> 

What I bemoan is the utter lack of hands-on experience. Most newly 
minted engineers can't even solder properly. Pathetic.


> Don't sell CS short.  It can be pretty grueling too.  We really 
> shouldn't be saying that one is harder than the other.
> 

Ok, peace :-)


> To get back on topic .....  We had cygwin on the lab computers.  
> I encouraged students to install it on their own computers if 
> they were running windows and didn't also have a unix-type OS 
> such as Linux, BSD, or Mac. 
> 

Maybe I should look into that as well. Although Sun VirtualBox does a 
fine job on mine if I want to run Linux programs (after using the 
ball-peen hammer on it a few times).

-- 
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.



_______________________________________________
geda-user mailing list
geda-user@xxxxxxxxxxxxxx
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user