OpenVSP 3.43.1 Released

[Rob McDonald]

One feature, a bunch of fixes, and lots of infrastructure in this release.

Spoiler alert -- 3.44.0 is going to be big and it is just around the corner. There is much to be excited about -- but it is so big that it is more risky than OpenVSP's normal releases. Because of this, 3.43.X will stay on the download page for a while and users will be encouraged to tread carefully with 3.44.X until things settle down.

With that in mind, there are a handful of fixes built up that have been hitting some users. Ergo, let me introduce 3.43.1.

First, the feature... You can now modify an airfoil section with a plain trailing edge flap. It is as easy as specifying the chordwise extent, hinge location, and deflection.

Next, there are a bunch of fixes. Watermarks weren't displaying. Calculix elements were numbered from 2 instead of 1. Multi-facade instances weren't starting when lots of parallel processes attempted to spool up lots of parallel facades. Yes, someone is doing that, why aren't you? FindThickness wasn't as robust as it should be. BOR XSecCurve Parms weren't updated from file load. Etc.

There were also some fixes for potential buffer overflows and a bunch of housecleaning found through static analysis. Use of std::map was converted to std::unordered_map wherever possible -- it should provide a theoretical speed improvement, but I doubt anyone will be able to measure any change in a real workflow.

Finally, GitHub deprecated their windows_2019 runners, forcing an upgrade to windows_2022. While this mostly went smoothly, it did cause a CMake race condition to appear. That was fun.

Overall, this is a nice bugfix release and everyone should update their production use of OpenVSP to this version.


Features:
- Plain flap airfoil modification

Library Updates:
- Update Code-Eli to fix FindThickness

Build System:
- Update GitHub Windows build to windows_2022 runner

Fixes:
- Fix large scale MultiFacade problem
- Fix Calculix element numbering
- Fix Watermark display
- Fix fscanf buffer overflows
- Fix FindThickness -- used by airfoil TE trimming by thickness
- Fix BOR file load

[Ганс]

Hi, I have updated to this version and it doesn't work with the older version files. For example, VSPAERO just closes OpenVSP. I've downloaded the 3.43.1 Windows for Pithon 3.11. Previously, I have been working on 3.43.0 version.

четвер, 3 липня 2025 р. о 10:00:07 UTC+3 Rob McDonald пише:

[Rob McDonald]

I appreciate the bug report, but you need to be a lot more specific.

What exactly happens, in what situation.  Best if you can include the file.  I need exact steps to replicate the problem -- and also a statement of what behavior you expected instead.

Rob

[Michael McCartney]

I have seen the same running Open vsp on Windows 10 and Python version 11.1.  Open VSP works just fine after I over ride Windows declaration that it is malicious software.  I am able to modify and save models created in version 3.43.0 using version 3.43.1. Version 3.43.1 will even read in previous  results generated by VSP panel solver in version 3.43.0. However version 3.43.1  VSP panel solver has a problem.  I can open the GUI form setup the analysis the exact same way as 3.43.0  then when I press start solver It crashes completely.  No warnings and any unsaved modifications are lost.   I have checked and cannot find any other files in the release that Windows 10 thinks are malicious software.   

"Scientist study the world as it is, Engineers create a world that never has been"
Theodor Von Karman

--
You received this message because you are subscribed to the Google Groups "OpenVSP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openvsp+u...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/openvsp/d1ce10f2-624c-4dc0-a589-1cb1d114870fn%40googlegroups.com.

[Michael McCartney]

I just sent out a detailed description of the problem I encountered here is the simplest of the files I tried it on.

Mike McCartney
316-204-8769

"Scientist study the world as it is, Engineers create a world that never has been"
Theodor Von Karman

--

You received this message because you are subscribed to the Google Groups "OpenVSP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openvsp+u...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/openvsp/1d7c91d8-198f-44a0-adb0-9ee4a8f586b1n%40googlegroups.com.

[Rob McDonald]

The malicious software warning is harmless.

Windows marks anything downloaded from the internet that is not 'signed' as malicious.

I spent some time a few years ago trying to get set up to sign our builds, but was unable to get through the process and bureaucracy.  Microsoft uses outside companies to do the signing -- they ensure you are 'trusted' by charging an exorbitant fee to sign binaries.  As an open source project, we can't afford to buy the trust.  There are some provisions for open source projects -- but they don't build trust in a project, they build trust in a single download.  Consequently, every time a new build is released, the first X downloaders would see it as untrusted -- if enough of them trust it, it would then become trusted for later people.  I tried going down the path of figuring this out and integrating it into our build system, but I was unable to get it all sorted.

We use an external library to perform constrained Delaunay triangulation.  This is notoriously tricky to get 100% robust.  A couple years ago, I switched from library A to library B in an attempt to improve things.  Unfortunately, it seemed to be causing more problems than it fixed.  In 3.43.1, I switched back to library A.  I suspect this switch is causing the problems you're seeing.

Thanks for sharing your file, I will try to use it to replicate the problem.

Rob

To view this discussion visit https://groups.google.com/d/msgid/openvsp/CADKkOzhTs5fqNomPhM3iVPTFRg59a0t8jA_8e8TCO6Wf4%3Du0kQ%40mail.gmail.com.

[Michael McCartney]

Thank you.  I agree you should not have to have this software signed by an outside organization.  I just remember that when I first tried to run open vsp 3.43.0 and 3.43.1 Windows prevented me from running until I used admin privileges to over ride the block. Version 3.43.0 ran fine after that until the first time I ran the vsp panel solver.  The first time I ran the panel solver Windows  forced me to override the block again.  With 3.43.1 It did not block me or give me the chance to override a block when I tried to run the Panel solver.  It just crashed.

Mike McCartney
316-204-8769

"Scientist study the world as it is, Engineers create a world that never has been"
Theodor Von Karman

To view this discussion visit https://groups.google.com/d/msgid/openvsp/CAEppYpH%3Dp3EBs2zmPXTvzTJO1HT%3D0AK4VCz5U08O4xCZaHynCQ%40mail.gmail.com.

[Rob McDonald]

This file seems to work fine on my machine.

Please tell me exact steps to replicate the problem with this file -- starting from opening the file.

I noticed that you saved the prior run's MeshGeom in the file (and had the main Geometry No-Show'd) while using the Shown Set.

So, I deleted the MeshGeom and Show'd the Wing before running.

Just to confirm, you're on Windows?  I am on MacOS (Aarch64) for most of my work.

Rob

[Michael McCartney]

Rob, I am on windows 10.  My machine does not support windows 11.0  

I copied that file into another directory.  Opened the file using 3.43.1.  Deleted the mesh.  Showed the geometry. Opened the solver gui, Launched the analysis.  Panel mode reference lengths based on model.  12 slices,  -4 to 15 degree incident in 20 steps. Jacobi preconditioner, 2d max Cl = 1.6,  

This wing is radically twisted to get a prandtl-d lift distribution.  I wanted to investigate what happens at higher angles of attack.  I was expecting to see the lift distribution change as individual section lift exceeded the 1.6 max 2D lift value.  Unfortunately it seems vsp aero is limiting the integrated lift CL to the max 2D lift of 1.6 instead of limiting the max section lift CLi to 1.6 as I was expecting.  Just the same it did run on my machine under 3.43.0  but does not run under 3.43.1.       

Mike McCartney
316-204-8769

"Scientist study the world as it is, Engineers create a world that never has been"
Theodor Von Karman

To view this discussion visit https://groups.google.com/d/msgid/openvsp/29095a48-877b-4395-a89d-ca1802d57fb1n%40googlegroups.com.

[Rob McDonald]

I'll have to fire up my Windows machine to see what is going on.

Here is the result of the cl distribution I get for the case as loaded from the file.

I'm not sure why it is limiting the cl to 1.8 instead of 1.6 -- perhaps it is something to do with the ratio of cbar to the local chord...

If you set the clmax to 1.422, you get something closer to what you're looking for...

The soon to be released new version of VSPAERO uses a fundamentally different way of limiting clmax, so you'll want to experiment with it when it is released.

I'm sure Dave will talk about that during the Workshop.

Rob

To view this discussion visit https://groups.google.com/d/msgid/openvsp/CADKkOzg54fav255fbb1buT1Ptt-Tv1qPO2YLdsaW%2B_3ke4Osow%40mail.gmail.com.

[Michael McCartney]

What you got is what I would expect.  I did not get this result in 3.43.0 run in windows 10 and it did not run on 3.43.1.  Looking forward to 3.44 hope it runs on my machine.  Looking forward to seeing the presentations either real time or archived. 

Mike McCartney
316-204-8769

"Scientist study the world as it is, Engineers create a world that never has been"
Theodor Von Karman

To view this discussion visit https://groups.google.com/d/msgid/openvsp/CAEppYpE3_uy0z_cyfVtKDeZqoJS_kPVZZoydbVmYJsMeBXaNFg%40mail.gmail.com.

[Rob McDonald]

Nothing has changed in the released version of the VSPAERO solver in a really long time -- all these versions should get the same results.

Make sure you're plotting cl -- the default is cl*c/cref -- which does not show the same visible flat-top.

Rob

[Michael McCartney]

Thank you I was plotting Cl*c/cref

Mike McCartney
316-204-8769

"Scientist study the world as it is, Engineers create a world that never has been"
Theodor Von Karman

To view this discussion visit https://groups.google.com/d/msgid/openvsp/ddf43bf2-3d64-4869-8e3c-ea7db482c803n%40googlegroups.com.

[Alex Sepnov]

Hi Rob,
it seems I got almost the same problem in windows 11. 
The file that I opened previously was saved file in OpenVSP 3.34.0. All the files generated during running the OpenVSP 3.34.0 were deleted, leaving the *.vsp3 only,

First of all, I opened the folder restriction for vsp.exe of OpenVSP 3.34.1, which is required as usual. The i opened the file, went directly to analysis > DegenGeom > then press 'execute'. Everytime i pressed 'execute', the vsp.exe crashed. I have no idea how to log the crash in windows 11, but i attach the file here. additional information, I am using OpenVSP-3.43.1-win64-Python3.11

alex

[Michael McCartney]

Rob,  First let me say I found the workshop to be very informative.  Thank you for all the hard work. I am looking forward to the enhancements in  VSP version 3.44.

I downloaded Alex's model onto my Windows 10 computer and repeated the experiment.  I got the same results.  It appears that executing both the vortex lattice solver and the panel solver result in an immediate crash of VSP.  I searched my computer for a windows crash log that would indicate what is happening.  I found nothing.  VSP 3.43.1 seems to just quit.  

Mike McCartney
316-204-8769

"Scientist study the world as it is, Engineers create a world that never has been"
Theodor Von Karman

To view this discussion visit https://groups.google.com/d/msgid/openvsp/57761ac4-ccfa-439b-8821-1cb8f1310107n%40googlegroups.com.