offset does not work well

[Anonymous]

I have a "mouse desk" on my chair arm.  The problem is that my cordless mouse suffers from severe depression and keeps jumping off onto the floor, trying to End It All, I guess. Finding it can be challenging; without a cord to tether it, it likes to skitter under the desk, under the bookshelf, etc.

So I decided to 3D print a "mouse fence".  To do this, I printed out a 1cm grid, taped it on, and cut it to fit.  Then I measured the distance between the edges (which I numbered).  I created a sketch which had lines constrained to these sizes and constrained to a common center line.  All is good so far.  I connected the endpoints of the lines with either straight lines or arcs to get the outline of the fence.  All is good.  I closed the loop.  Still doing fine.  I used "Offset" to create a 5mm-wide inner fence.  A bit of a pain because there did not seem to be a way to specify the offset while doing it.  But I figured out that I could use "Dimension" to specify a 5mm separation.  Only the bottom half of the offset lines redimensioned.  I tried making a coincident point between the top offset and the bottom offset, but instead of dragging the point over to match, it created some weird self-intersecting zigzag patter.  I deleted this, and specified a 5mm dimension for the top half of the offset.  OK, still good.  I then created the coincident point at each end (left and right) of where the two halves met.  They were cut in the center of the XZ plane, which was visible when I did this.   Well, annoying, but I've got it.  I have already closed the outer loop; I now close the inner loop.  This seems to go well.  I close the sketch (Sketch 3).  Then I select "Extrude", and I do not get the fence shown in the attached file.  I get the entire top half being extruded.  Same for the bottom half.  Yet I have two closed loops and it should let me select just that space.  I struggle with this, no luck.  I give up, and create a new sketch, Sketch 4.  I then project geometry of each of the two loops from sketch 3 into sketch 4.  I exit sketch 4, mark sketch 3 as invisible, and return to sketch 4, only to discover that I had missed a few lines.  Back and forth a couple times to fix these.  (Why can't I just reach over when editing Sketch 4, select Sketch 3, right-click, and change its visibility?  This is not the first time I have wanted this feature).  OK, I get all the geometry projected.  Now, in Sketch 4, I close the outer loop.  This works.  I close the inner loop.  This works.  I exit sketch 4.  (Remember that Sketch 3 is still invisible).  I select "Extrude" and click between the inner and outer loop, as I did with Sketch 3.  It works perfectly.  I take the default 1cm extrusion.  I want the top to have a half-round fillet.  I select Fillet, select half-round, and it sets me up to select face set 1.  I click on all the outer faces; select middle face set, click on all 26 top faces, select face set 2, click on all inner faces.  Click OK, and it tells me it failed to create the fillet.  Why?  Another one of those Inventor secrets.  The person who wrote the code knew what was wrong, but refuses to tell me.  I get the generic message.  This has all the charm of a compiler that finishes up the compilation and issues the only error message the compiler knows about:  "You have syntax errors in your file". Not helpful.  OK, I give up.  I create a 2.5mm edge fillet on the outside edge; this works.  I create a 2.5mm edge fillet on the inside edge.  This works.  I color the edges so they can easily be seen (Oh, for "Select face only" as a selection option!), and I am done.

So I am left with questions:

• Why didn't it let me select the area between the two (closed loop) outlines in Sketch 3?
• Why did it give me a top half and bottom half as separate creations when I did the offset?
• Why did it not drag the top offset to match the bottom instead of creating a convoluted connection?
• Why is it so hard to project geometries?  (See my Inventor Idea on closing loops)
• Why would it not create a half-round fillet?
• Why did it not tell me exactly why it would not create a half-round fillet?
• Why was it necessary to create Sketch 4, which worked exactly like I expected Sketch 3 to work?
• Why couldn't I click on the 3D model in Inventor (on my left screen) and just drag it onto the "drop file" box?

[JDMather]

---

@Anonymous wrote:

So I am left with questions:

• Why ... in Sketch 3?
• Why was it necessary to create Sketch 4, which worked exactly like I expected Sketch 3 to work?

---

No file Attached???

[Anonymous]

I distinctly remember bringing up file explorer, locating the file, and dropping it into the place on the Web page where a file was supposed to be dropped.  Silly me, I thought this would have attached the file.  This time, I used the "browse" feature.  And I checked to see that the file was actually attached.

[Anonymous]

I just tried the drag-and-drop, and although it appears to have taken the file, the file did not appear as an attachment.  This is clearly a bug in the Web page.  I didn't bother to send the test message, because I already sent the file successfully (or so it claims).

[melrosecad · Accepted solution]

---

So I am left with questions:

• Why didn't it let me select the area between the two (closed loop) outlines in Sketch 3? You have a line splitting it up. Change it to construction and it works.
• Why did it give me a top half and bottom half as separate creations when I did the offset? Again you have a line splitting up the geometry.
• Why did it not drag the top offset to match the bottom instead of creating a convoluted connection? Works fine with the said line being made construction.
• Why is it so hard to project geometries?  (See my Inventor Idea on closing loops) There does seem to be some weird inconsistencies with viability of sketches you want on/off when working in another sketch. It annoys me too.
• Why would it not create a half-round fillet? If the edges were continuous it would likely work. 
• Why did it not tell me exactly why it would not create a half-round fillet? This is an unrealistic expectation for software to know exactly why every error occurs.
• Why was it necessary to create Sketch 4, which worked exactly like I expected Sketch 3 to work? See above about the line.
• Why couldn't I click on the 3D model in Inventor (on my left screen) and just drag it onto the "drop file" box? No idea.

[Anonymous]

Thank you for the answer.  It clears up a mystery that should not have been a mystery.

I do want to comment on the "unrealistic expectation" comment. I am shortly starting my 58th year as a programmer, and in that time I have learned that every condition that issues a message has a "root cause". There is a line of code that says, somewhere,

if(something)
      return false; // unable to create fillet 

and someplace else there is a statement

if(other_condition)
    return false; // unable to create fillet 

In my world, I would either throw a specific exception or I would not return "false" but an error code. So I might write this as

if(something)
     return ERROR_TOO_MUCH_DETAIL;
...
if(other_condition)
     return ERROR_DISCONTIGUOUS_SEGMENTS; 

so don't try to explain to me that it is "unrealistic". I have found that doing it this way has several advantages:

• The user can often determine, based on the error message, what has gone wrong
• Tech support costs are lower because there is not a generic "it didn't work" message that lumps all failures into a one-message-fits-all universe. The tech support teams at my clients have all these error messages indexed (something I get free from my toolchain, because I have arranged that it be so), so they can quickly suggest the remedy. Some error codes say "This error message represents a response to an impossible condition", and that's when I get an email.
• My response to my clients is rapid, because I can uniquely identify the exact line in the source that issued that error message. I never use one message when six messages make it clearer. Which means I have a better chance of getting rapid turnaround on a point release.

I simply do not believe the statement that the expectation is unrealistic. It is very realistic. I was doing this forty-five years ago, and it has always paid off. I used a database system in which every error code and every error text appeared, and a paragraph of explanation. I would get a header file with the error codes and their associated numbers, the text of the message would go to a file that the program could quickly access (we were working on minicomputers in those days), and the error code, its number, and the paragraph of text would appear in a printed manual. The error codes were sorted alphabetically, and there was an error-number-to-name map as well. All it takes is a commitment to make the UX as good as possible.

The nice thing is that for Windows apps, Microsoft has a tool that does this; I predated that tool by about 20 years. So I know that every error has an if-statement that generates it, and every if-statement has a context that it is working in. I want to know that context. I want to, when I get a particular error text, or code, such as "xxxxx", to "google xxxxx" and get an explanation, preferably from a document on the Autodesk site. It can be done. It is important for lots of reasons. And it is far, far from "unrealistic".

You were able to identify the error. But I should have been given sufficient information to figure it out for myself. I did not see the line because I did not know to look for a line.

[melrosecad]

My comment on that was only based on my experience with using software of all different types over the years where it seems to be a common theme that error messages are vague. I am not a programmer but can only imagine it's not something simple or worth the development time to solve or it wouldn't be such a common issue. The reasons for why a fillet might fail could be numerous.

As for being given information about looking for that line. That comes from basic knowledge about using the software and general best practices most which come from experience. 

[Anonymous]

Yes, as a new user I am at a distinct disadvantage here. Which is why I need all the help I can get from the program.

I believe the reason error messages are obscure is that programmers don’t see the value in making them clear. I would be embarrassed to deliver a product with messages this bad. I don’t think I’ve ever written a program that had bad error messages since sometime in the early 1970s. And generally it doesn’t get priority because either the tech support budget comes from somebody else’s bottom line (never mind that it raises the tech support cost to the company by a factor of 10), or the programmers are just lazy, or have been taught bad style. If there are 200 reasons a fillet can’t complete, I would have had 200 uniquely-identifiable messages that explained what went wrong. They might have been in the form:
Fillet could not be completed (INV433)
But that 433 is the important part of that line; each abort has an error code that is unique.

When I teach courses (which I did for 25 years before I got sidelined by health issues), I would ask the students,
“Which error message would you like to see?” and the choices were

File Open Error

And
File Open Error
C:\user\appdata\startup.cfg
Error code 2 (file not found)

So they call up tech support, and it costs $50 just to answer the phone. If tech support can’t figure out what the first message means, there is a serious problem. It might even escalate back to the developers, at which point the cost to the company is now hovering dangerously close to $1000. In the second case, the user says, “Oh, crap, I deleted the old file and forgot to change my project settings” or words to that effect, and fixes it. Or tech support knows what went wrong, and asks, “Did you…” and the user says, “Yes, but how do I fix it?” and tech support says, “You…” and five minutes later the call is closed and the user is happy. If I get “Fillet could not be completed (INV433)” I go to the help and type in INV433 and up comes a page of explanation of why that happened and what I can do about it. If it doesn’t, I post it here, some nice person solves the problem for me, and it has cost all of us time and effort. But if I see INV433 come up again, I will know this time what went wrong. And I can fix it. Nobody spends a lot of time with it, and everybody is happier. It isn’t that hard to do. Really. I made my living (successfully, I might add) doing exactly this, and I won, my clients won, and their customers won. Not often you can get win-win-win. What is frustrating about Inventor is that I keep finding myself in the “lose” category. If it weren’t for COVID-19, I would probably be spending a huge pile of money to get local Inventor hands-on training. I had already decided to do this when the world fell in, and I learned a new word: “comorbidities”, kind of important, because I have five of them. I have not been out of the house for more than five months.