Repository navigation
release v8.4.1 (or v8.5.0) #366
Description
EDIT: i removed original message an just keep track of ongoing actions
Proposed changes for v8.4.1
-Fix profile plot temporarily not updating after subtracting or averaging wavefronts (#343 done in Februrary)
-Fix build issue with Clang compiler (#347)
-Fix outline helper dialog not closing when the application exits (#352)
-Fix minor UI issue on the 16 diameters profile plot (#355)
-Fix potential dip in profile plot when using Gaussian blur + minor display issues (#359)
-Fix very minor gui issue - save file dialog says foucault instead of ronchi when saving ronchi (#359)
-Fix scroll did not update min/max values in countour plot (#359)
-Fix some of the dialog not closing on app close (#360)
-Fix Gaussian blur checkbox not displaying actually applied settings (#361)
-Fix central hole and regions were not cleaned from previous OLN file (#362)
-Fix elliptical flat configuration bug and ensure minor axis is always smaller (#364)
-Fix import wft config yes button was not working at all (#371)
-Build pipeline now publishes a .dmg package for macOS users (#348 )
-Build pipeline now publishes an .AppImage package for Linux users (#365)
-3D view interracts better with mouse. Left click orbit, right click pan (#381 + #382)
-IMPROVE Zernike enable table is saved for each wft. Display diff (#369)
Ready for review
In progress
-[NOT READY] fix wrong zernike fit in case of (annular obstruction AND aperture reduction) (#372)
-[not ready for review] fix mirror settings modified in registry before user press accept button (#370)
-Reviewed wavefront code (#356)
-Fix ignore edge region. Center edge had wrong size and could have wrong coordinates (#375)
Activity
gr5 commentedon Aug 7, 2026
Fix profile plot showing the wrong plot after subtractionChange to:
Fix profile plot temporarily not updating after subtracting or averaging wavefronts
Get rid of quote after the 16. You can just say 16 diameters. Or diameters at 16 angles.
gr5 commentedon Aug 7, 2026
Your proposal is fine with me but I don't think there is a rush. I'll be back home with more free time in 8 days. I'm actually testing PR #359 at the moment.
I agree with the basic sentiment that we don't need a ton of QA before we create a release. It's just that maybe it's better to do one release than 2 releases a week apart.
Lucassifoni commentedon Aug 7, 2026
Hi,
I am happy to finish the two PRs but my availability is spotty at best right now since I am on holiday. My next moment of contiguous availability looks to be similar to gr5's in ~7-8 days.
atsju commentedon Aug 7, 2026
Sorry I din't want to insist on the fact it needs to be done fast. We can do the release later, even if it's in a month or two I'm fine with it.
I was just thinking we can do a release for the Linux/Mac first as this might be very valuable for some users.
And then have a ton of bug fix and features before next release. With all the fixes it will probably be some time before we are ready to release again. I hope this clarifies.
gr5 commentedon Aug 7, 2026
All the PRs that you think are ready will be tested by me within 10 days or so.
gr5 commentedon Aug 7, 2026
pr #359 is particularly difficult as it has the potential to cause difficult to notice bugs with subtraction and averaging features. Perhaps something else as well - oh yeah - ronchi and foucault feature could be affected as well.
atsju commentedon Aug 7, 2026
I will let you manage the merge and order. I'm fine with not doing the intermediate release if you prefer to have one larger release.
githubdoe commentedon Aug 7, 2026
I'm surprised by these. They work on what I'm using.
Fix profile plot temporarily not updating after subtracting or averaging wavefronts
Fix minor UI issue on the 16 diameters profile plot
There is no longer a 16 diameters profile plot. It is just one of the right pop up menus on the profile. When selected it displays with any other profile plot also selected.
githubdoe commentedon Aug 7, 2026
The current profile plot should look like this. Does the one your using? I'm currently using DFTFringe and do not want to risk loosing my work to start up that latest release. I thought the latest release looked like this.
gr5 commentedon Aug 7, 2026
Dale if you want to see the issue, it's described here from a guy named Zack Steinburg who I've coincidentally met in person.
#345
There's a 17 second video showing the bug. It's extremely minor. Anyway @atsju fixed it. Or you can just read the text how to duplicate the issue.
gr5 commentedon Aug 7, 2026
Maybe you fixed this one but didn't check it in. The other 2 plots update immediately but the profile plot doesn't update instantly. You have to do one of a few dozen things to get it to reload. For example, click on the wavefront. Or click recalculate. Or change the gaussian blur. Or one of many other ways to get it to reload (e.g. right click on it and change the settings).
gr5 commentedon Aug 7, 2026
With this big flurry of activity it may seem like there are major changes but there are not. Most of the activity are not even code changes but changes in the build process for mac and linux. Most of the PRs are only a few lines of code.
gr5 commentedon Aug 7, 2026
Oh. And yes, that's what the popup menu looks like.
githubdoe commentedon Aug 7, 2026
I hope the profile also looks like that. There are less buttons on the top than older versions.
githubdoe commentedon Aug 7, 2026
When I get some time I will diff what I have against the master. Currently Qt says I'm on a project called master. But at one time I had a branch called master on my machine I think that was not really the remote master. I get so confused by github as you all know.
24 remaining items
Lucassifoni commentedon Oct 28, 2026
A friend reports the same crash on Linux when going in the video source selection panel. But of course with a different cause. I'll try it on linux too.
githubdoe commentedon Oct 28, 2026
I down loaded the latest version. The released version fails to connect to my URL link. I have yet to merge to my archive. Once done I will try to debug but that may be awhile.
gr5 commentedon Oct 28, 2026
If you are using the blinkrate branch, please merge master into it before making edits. Don't forget to switch to master, pull that, then switch back to the branch you will be working on before merging.
I'm quite surprised the web stream would not work as it was working for you recently.
gr5 commentedon Oct 28, 2026
I have a raspberry pi and an hq camera. Is it possible to turn that into a web streamer so that next time I can test it as part of the QA process?
google AI says I can use mediaTMX. Is that what you use Dale? I can try that when I'm back home in a few days.
githubdoe commentedon Oct 29, 2026
Just down load my skysolve. Then it is all ready to go. Make sure you get version 2. It is on github.
githubdoe commentedon Oct 29, 2026
There is no way in heck that I will merge now. I need a working version and I have one. Instead I will do what is necessary to have two local source stacks for debug after I get some time. But first I will look at the camera code to see where it might have broken. But first I have to get the source without destroying my local branch. So I will do that first.
githubdoe commentedon Oct 29, 2026
Skysolve will take images and try to match them to the sky. To keep the solve from slowing it down click on the align tab on it's gui. It will do a most 1 more solve after that then stop. But it will continue to grab images as fast as it can and stream them out.
gr5 commentedon Oct 29, 2026
Well you could for example just get a diff between your current code and the master on github by doing:
git fetch <-- that just makes sure your local repository knows about all the changes to all the branches but doesn't move your branch.
Then you can google how to diff the current code with origin/master. And just look at the files related to video.
Or, assuming you have everything you wrote committed to some commit somewhere...
do a git stat and make sure everything is committed.
Then see what version is your current commit. I'm guessing this one on 10/08 as that is your most recent commit that you pushed to github:
"disabled loop timer display" was your comment and the hex sha starts with: 153f3c8
You can always get back to it with: git checkout 153f3c8
google ai can help you move around between commits so you can try different versions and see where things went wrong and always get back to where you are now.
No need to have multiple repositories.
gr5 commentedon Oct 29, 2026
I can't try skysolve until I'm back home where my rasp pi is, which will be Saturday.
By the way, have you heard of astrohelm? I think it does the same thing as skysolve but with a smart phone. It's supposedly extremely accurate even though the camera is typically a pretty wide field. Just strap it to your telescope aligned with where the scope is looking and it plate solves for you. Just like skysolve.
githubdoe commentedon Oct 29, 2026
There are several EFinders available these days. The problem with strapping a phone to your scope is that you need to see the screen to see what skysolve is telling you or any app you care to use. It is much better to have your phone in your hand or in my case I use a tablet on a stalk.
You will find the image too bright on the GUI of skysolve for an indoor image. There is a tab called "Norm" (normal brightness) that you should press so that skysolve does not brighten the cameras image.
githubdoe commentedon Oct 29, 2026
I have tried in the past to use GITHUBS magic words and commit ids and most time I destroyed my local code. No more. Multiple archives work just fine. AI even recommends that as the safest way.
githubdoe commentedon Oct 29, 2026
Another thing about skysolve. It needs it's own SD card. It would be tricky to put it on an existing card because at bootup it will try to make a wifi hot spot. So you can either learn about how to use it. Or get AI to write you a streaming services and also make a wifi hot spot for it. Either way I think you have to make a hot spot and the connect your laptop to it. If you do use skysolve then you also have a neat tool to observe with.
githubdoe commentedon Oct 29, 2026
I did find the issue. Well AI helped. For some reason the open statement that opens only URL's needs another parameter. Why now I do not know. But it works. So in the video worker class line 34. This tells it to use FFMPEG which of course might be an issue if the user does not have FFMPEG installed and known by the system. I do not know if that means streams only works with FFMPEG but I think so.
instead of
m_cap.open(m_source.toStdString());
gr5 commentedon Oct 29, 2026
gemini says: No, you generally do not need to install FFmpeg separately on your system if you are using pre-built OpenCV binaries
Also if you leave that out, opencv is supposed to cycle through a bunch of options but choosing this sounds like a good idea overall as it is the "gold standard" of streaming video and selecting this will usually cause less issues regarding dropped frames, etc.
githubdoe commentedon Oct 29, 2026
Yes that makes sense.