8 Replies Latest reply on Aug 14, 2012 3:01 PM by Paul Passarelli

    Copying 3d sketch causes "General Protection Fault"

    Paul Passarelli

      I just finished rendering a set of sketches for the lube oil in my SLDASM, and hit save.

       

      Then I started to make them into parts.  I managed to do this 3 times successfully, yesterday.  Today I Inserted New Part, hit copy on a 3D sketch in the assy and hit paste into the part -- nothing happened...  that's to say the UI froze.  I alt-tabbed away then came back to a solid black work area and a still frozen UI.  Several minutes later the Unexpected error dialog box showed up, and --

       

      Once again, my work from the *ENTIRE SESSION* has been bolled back into the bit bucket!!!

       

      If anyone at SolidWorks Corporate is reading this... I want this stability nonsense fixed.

        • Re: Copying 3d sketch causes "General Protection Fault"
          Alin Vargatu

          Paul, I am a bit confused. You said you hit "save". So how come you lost the whole session?

            • Re: Copying 3d sketch causes "General Protection Fault"
              Paul Passarelli

              Hi Alin, 

               

                Confusion is understandable.  Some *BUG* caused the s'ware to crash (fault).  I can deal with bugs, even bugs that cause faults.  What I WILL NOT TOLERATE is incautious algorithms that allow the s'ware to destroy my work -- NO! That's not quite it -- The s'ware has an "AutoBackup" setting that steals 2 minutes out of every 15, and thanother setting that prompts me to save every 30 minutes which takes up to 5 minutes when it decides to rebuild everything...  and THEN my work STILL gets corrupted!!!  That's what has me red-faced-fuming angry.

               

                I to am confused as to why the s'ware would leave files in an uncommitted state. -- this is an inexcusable state of affairs!

               

                It is a symptom of "non-comprehended complexity", in the algorithms that actually execute tasks.  My actions were *never* committed to my files, sure they were swapped, cached, _memset(), etc...  but never *properly* committed.  It's a logical bug in the s'ware. 

               

                So, when the fault occurred, the OS (Win 7 Pro NTFS), in an effort to preserve filesystem integrity, simply unwound the offending application's pending writes to disk.  Et voila! -- everything was rolled back to the way it was when I started the application!

               

                IMHO, the s'ware is built on a *very shakey* set of core routines.  That's really all there is to it.  Sure they've spent lots of time & resources knitting together all types of cool features, but it's a tower built on sand.  I would guess that developers are frequently making "end-runs" around the API's, because certain data structures are not available under the conditions established.  It's like painting one's self into a corner and deciding that one or two footprints in the paint are "acceptable"...  There is an addage that there is never time to do it right the first time, but there is always time to do it again to make it work.  Well, it's way past time for SolidWorks fo find the time to do it right!!!

               

              Respectfully,

                • Re: Copying 3d sketch causes "General Protection Fault"
                  Alin Vargatu

                  Well, if you save a file, the file is saved on the disk (unless you dismissed a warning message some time before).

                   

                  Check your dismissed messages.

                    • Re: Copying 3d sketch causes "General Protection Fault"
                      Paul Passarelli

                      You and I would *think so*; but whatever remains is the truth! 

                       

                        When I say files are not committed, I'm drawing an analogy with a database paradigm for guaranteed data integrity. 

                       

                        A "transaction" is begun, the data files are "opened, read, accessed, and finally written", but nothing is permanent until the transaction is "committed" and the committment operation is completed successfully.

                       

                        The paradign is computationally strong, and was developed to secure financial transactions...  What I suspect is that the s'ware has the skeleton of secure data handling, but that when the fault occurred, it was caught by a handler that was unaware of some global semaphor or mutex, and so the fault handler was unable to properly unwind my work.  Unce the application thread died, the Windoze did *exactly* what it had to do to preserve filesystem integrity.

                       

                        As for warning messages... it's irrelevant.  I saved "successfully" or so it seemed, or rather so I was told...  The only way I seem to be able to *KNOW* that my work is safe is to exit SolidWorks, wait for its threads to go *completely* away (otherwise I get the "unable to open journal" error), and reboot when the application simply hangs on start.  BTW SolidWorks is the only application I own that will refuse to start promptly unless it can "phone home" after several restarts.

                        • Re: Copying 3d sketch causes "General Protection Fault"
                          Kelvin Lamport

                          Are you still running with PAE set?

                            • Re: Copying 3d sketch causes "General Protection Fault"
                              Paul Passarelli

                              Hi Kelvin,

                               

                                Um, I'd haf'ta say 'dunno, 'prolly not, but maybe?  :^)

                               

                                PAE (Physical Address Extension) is a capability of the chipset.  If I were runnin a linux kernel, that I'de be able to say with certainty if I was running a PAE kernel.  But since I'm running Windows 7 Professional 64-bit (retail) out of the box, a cannot say for certain -- I can say I didn't expicitly activate and registry settings that explicitly refer to PAE.

                            • Re: Copying 3d sketch causes "General Protection Fault"
                              Alin Vargatu

                              Paul Passarelli wrote:

                               

                                The paradign is computationally strong, and was developed to secure financial transactions...  What I suspect is that the s'ware has the skeleton of secure data handling, but that when the fault occurred, it was caught by a handler that was unaware of some global semaphor or mutex, and so the fault handler was unable to properly unwind my work.  Unce the application thread died, the Windoze did *exactly* what it had to do to preserve filesystem integrity.

                               

                               

                              Paul, thanks a lot for the detailed info provided. One more question for you as a software coder. Does the fact that I have never encounter these kind of issues, mean that my system is better set-up for catching these commitments, or should I be afraid that what happens to you on a regular basis might happen to me in sometimes the future?

                                • Re: Copying 3d sketch causes "General Protection Fault"
                                  Paul Passarelli

                                  Hi Alin,

                                   

                                    "Be afraid. Be very afraid." -- Gina Davis in "The Fly"

                                   

                                    What I believe is that if I were to sit down at your workstation, it would only be a matter of hours before your system started exhibiting the exact same bad behavior as my system.  Perhaps I'm cursed. ;^)

                                   

                                    I realize it sounds silly, but when I am working at the speed of thought the crashes are the hurdles that constantly slow me down.  When I'm killing time and something else is occupuing my mind, say I have my video editor up and I'm only playing SolidWorks while I'm watching my recordings and leave it alone while editing out the commercialss, it doesn't seem to kill me, nearly as often.  Then again, I'm generally not doing anything really important during that time.

                                   

                                    But when I'm *working*, then SolidWorks is getting 100% of my attention, and I expect 100% of it's ability.

                                   

                                  BTW, last Saturday I added 4G of Ready Boost memory, interestingly, my Windows Experience score dropped by 0.1 point (pfft), yet anecdotally, the performance of everything *except* solidworks seems to have improved a tiny bit (otherwise imperceptible).  OK maybe SolidWorks starts a tiny bit faster too, but the errors are still there, witness my thread from today, and a problem with Feature Tree folders.  I had been hopeful after yesterday, but confidence is down again today.