The “Good Ole Days” – Smackin’ Bugs
The “Good Ole Days” with SolidWorks means the days when all we had to complain about was bugs in the software. Bugs were at least something you stood a chance against. Corporate insanity is a different beast, and while I enjoy mocking it, I’m not vain enough to think I can have any effect on it.
So today was one of those days I ran into several regressions 2012 – 2013. When people ask you if they should change to a new version, you can never give them a really good answer until you know about the regressions. And you never know about them until you’ve been through every square inch of the software forwards and backwards a couple of times.
Here are two. Maybe you can add some that you’ve found.
First, I have a model built in 2012 with a Split feature, where a sketch splits the part. But if you edit it and recreate it in 2013, the sketch will only split in one direction, and NOT the direction it originally split in. This was a model that had a “black hole” feature in it that I talked about some time ago. Still has that anomaly if you cut it in just the right place.
Next, if you CtrlQ the model in 2013, one of the Fill features fails because somehow SolidWorks forgot a couple of edges in the selection. From there just about everything fails, including trims flipping.
Ok, so that means repairing a half a dozen features, including some wigged out offset sketches. Whatever. You get used to this kind of thing when you use SolidWorks for this kind of work.
I’d like to point out though, that this brand of making you rework your models is the domain of history-based models. It just doesn’t happen in direct modeling. Direct modeling doesn’t re-calculate stuff that you don’t want to change. Now granted, there is a difference between history-based swoopy surfaces and direct edit swoopy surfaces. I’m so tired of losing time like this because somebody recompiled the software on a Tuesday instead of a Wednesday. I really do lose a lot of time to this kind of problem. Sending a bill to Concord -er excuse me – Waltham doesn’t seem to do any good either.
Anyway, the second regression is one of those basic things that you learned in your first week with the software, and you oogled “ooooh, aaaahhh!” because SolidWorks worked just like Windows. If you window select several parts in Windows Explorer and drag them into an assembly, they are supposed to all pop right in, right? Well, not in 2013. I tested it in 2012, and it still works there. Can anybody confirm this one? It just puts the part into the assembly that you had your mouse over. Remember that universal joint crank assembly you do in the basic training class? That one used this method, or at least I did when I taught it.
I’ve paid for subscription service, but I gave up reporting bugs long ago. I’ve found that reporting them here in public is much more effective than allowing them to keep the warts private. Plus, it allows old what’s-his-name to keep padding his statistics without the added stress of reality.
We all want controllable reliable geometry. The butt crack and hogback type of defect is almost certain for Solidworks boundary surface and lofts. It is possible to reduce these with careful adjustment of the profiles and guide curves. They get smaller but are still present. Parasolid is not the problem. I really doubt that it will be more simple to write a new modeling kernel and a cad interface than to do the math properly to define better shapes.
One of my pet peeves is the size of Solidworks files. These are at least 1200x larger than the information to define the shapes. The files contain a bunch of wasted bytes, every deleted feature, history about every version of SW that has touched the file, history about where the file has been, basic spyware functions. I counted the keystrokes and mouse moves to generate an actual part, and compared it to the SW file. A dramatic example is to draw a helical spring. It is basically select a plane, draw a circle, select another plane draw a circle, generate a helix from one of the circles set the phasing, sweep the second circle. It takes less than 100 characters and mouse drag and clicks, the file will be about 1MB.
@AB
AB in case you look for some input, read this old article
http://soliddna.wordpress.com/surfacing-with-sewst-grand-grand-mother-of-st/
This one is more recent to be prepare for ST6
http://www.soliddna.com/SEcommunity/blog/1/entry-17-get-prepare-for-st6-surface-design/
If you want to read more sing-in for free
I don’t think its necessarily a poor implementation of parasolid its just that SW have tuned it to give the best possible general solution in the most number of cases. That is, its somewhat more tolerant and flexible than especially accurate and as such slightly lax in regard of the normal to condition. In the instance where conics are the basis and perhaps the user has a different purpose than achieving swoopiness it doesn’t offer the results we would prefer. I think the code can be targeted/optimised to please more users and considering the type of work they do. I note Mark hasn’t commented on this matter but perhaps that’s because SW don’t want to commit to anything given the near end game for SW. I don’t imagine Mark being the good guy he is and advocate for continuous improvement would not want to do something about it if he could. We are quite a demanding audience though. Possibly if we are all good little boys and girls something will turn up under the guise of what they term a ‘delighter’ 😉
Thanks for the answers.
@AB
AB it is not a PARASOLID issue at all.
It is an issue of how SW utilizes the kernel and it is there resultant math that defines the surfaces.
Ricks $1200 dollar add on that generates you high quality surfaces utilize the PARASOLID kernel also. They just do a more refined set of resultant math to define their surfaces.
So does NX and SE and a host of other software programs.
It is SW own implimentation of how they define and generate surfaces.
@AB
No, this is not a Parasolid problem. If it were we wouldn’t have tha capabilities in NX Surfacing or any other decent surfacing package using Parasolid.
What it is a poor implementation of code that uses Parasolid. My personal opinion is that this poorly implmented and written code is one of the key factors that drove DS to develop and new product based on their kernel. If you have to start over you might as well start over using your own tools (kernel).
@Rick McWilliams
Rick,
I too have experienced SW surfacing “undocumented” features myself. And am aware of the many animal references U bring to everyones attention.
To me it seams that the much ballyhoo’ed SW surfacing package is not at all what it has been played up to be.
If your $1200 dollar add on utilizing the PARASOLID kernal generates a higher quality surface that what comes out of the box is SW. To me dictates that the SW implimentation of surfacing utilizing the PARASOLID kernal is not very good. However they have made an effort to have many ez buttons and enhancements to user interface etc to give lots of feedback when surfacing.
I have put up with imperfections in the SW surfacing … but i’m dealing with small stuff. Not like you generating surfaces that are representing full scale airfoils.
Right now im working a new project. Small portable device. I’m going to be using ST5 sync sketches and ordered surfs, will let you know how it goes.
Also after thats done I should be able to try some of your airfoil stuff. And I am eagerly awaiting ST6 also to see what they have done with it.
Billy
@Rick McWilliams
>I really like discoverable geometry. I do a loft and discover that it is all wiggly and twisted, add a few guide curves and discover it is untwisted but still slightly deformed. Never do they say what the shaped between sections are supposed to be, or when it will ignore sections.
Isn’t it a parasolid kernel problem ? In this case, this would be an important reason for Solidworks going to catia kernel ?
@ralphg
Ralph, I agree this move seems deliberate and petty, but it’s funny how you always overlook the other side of the coin.
http://www.winnipegfreepress.com/canada/fantinos-office-claims-document-on-english-only-correspondence-was-altered-202204611.html
(I have zero trust in Fantino’s rebuttal or the actual federal government)
But in the end I would have prefered you refrained from tainting this blog with politics that no one else here cares about.
@Marc Gibeault
It is easy to write, “Not true.” It is easier, however, to provide proof:
http://fullcomment.nationalpost.com/2013/04/04/chris-selley-pauline-marois-pursues-a-sovereign-grievance-based-quebec/
In French: http://www.journaldemontreal.com/2013/04/02/cest-en-francaisque-ca-se-passera
These red threes are awful memories… Trying to remember what a feature was supposed to do in a particular model, re-selecting trimmed surfaces again and again.
It happened day after day, version after version. It was worse as the time passed by because our models and assemblies were getting more elaborate.
I’m now free of this stress; I model high-level stuff in Rhino and an army of engineers do the detailing in CATIA…
Not true
@Rick McWilliams
Rick what does that addon cost nowadays
Ryan,
The butt cracks and hogbacks are part of the character of a pig. Kind of cute in a piggy sort of way. They are somewhat less cute on the fuselage of an airplane or the roof of a car. Solidworks does not like to generate class A surfaces with controlled inflection lines. These defects are sometimes subtle but never desirable. I now use GW3D surfaces instead of SW surfaces. Add on cost about $1200 worth it for me.
In 2012, I used to be able to ctrl+copy a drawing in AutoCAD 2012 model space, and paste into a SolidWorks sketch. That doesn’t seem to work anymore. Either it’s SolidWorks 2013 that is at fault, or AutoCAD 2013, not sure since I updated both.
Rick,
With shapes like that, I’m wondering if you are modeling a sow pig? Hogbacks, curls, butt cracks, etc. 😉
Ryan
I really like discoverable geometry. I do a loft and discover that it is all wiggly and twisted, add a few guide curves and discover it is untwisted but still slightly deformed. Never do they say what the shaped between sections are supposed to be, or when it will ignore sections. We get to discover that splines get uptight, but what is the difference? What is going on when a spline handle changes another spline handle? If I wanted to move the other handle I woud have grabbed it. I feel so leading edge when I discover that most boundary surfaces produce butt cracks, hogbacks and sometimes tits, wrinkles and curls.
I used to practice vocabulary when I encountered a bug. I feel that I paid for professional software that should not have so many conspicuous bugs. I just do not know enough languages to have enough four letter words. It does not really make me feel better. Now I just make an entry in a document of bugs. I only count bugs encountered when trying to avoid bugs doine real work. I know how to make almost every type of feature fail or produce bad geometry.
How many years old does a bug need to be to be considered a classic? Do they really not care?
They never have tested their software thoroughly prior to release. I’ve been shouting about this for over 14 years. Has anything changed? Nope.
You see their business model says the USERS will test their software and report bugs for FREE. Get it? Think of all the money they’re saving.
And even better they charge you, the USER, for their mistakes, yeah really. It’s called Subscription Maintenance.
That way they can put out sloppy work, make mistakes, and charge their customers for the Service Packs that fix THEIR mistakes. Brilliant!
Way to go! Yeah, stick it to ‘um! Double Dippin’ The “Corporate Way” Money before anything else…
@TOP
Paul,
I hope you are being careful, given the situation where you are. Of all the places you could have picked at this very moment…
Best of luck, and take care.
@TOP
Not so far out of line. In the Canadian province of Quebec, the government last week ordered its bureaucrats to speak with others in other provinces in French only. If the others don’t understand French (and most wouldn’t), then the Quebec bureaucrats are prohibited from speaking English, and a translator must be hired, and paid for by tax payers. No joke.
In dealing with corporate insanity you Will have to blog in French now.
IN THE GOOD OLD DAYS HUMANS TESTED THE SOFTWARE. They could see bugs that automated testing can’t because it ca’t remember the past. FWIW does this type of pro let reflect the level of bug testing in CATIA?