×

INDI Library v2.0.6 is Released (02 Feb 2024)

Bi-monthly release with minor bug fixes and improvements

Issues with latest build

  • Posts: 1009
  • Thank you received: 133
I had an observing run with the latest git version of kstars (kstars-3.5.4-2157_g7498a113d, self compiled)
While stability had improved (3.5.4-2110_ga4b519ad1 was crashing repeatedly while just guiding), this one still died during focusing after the second filter change (which unfortunately costed me 50% of the rare good night   ).
It died with a segfault, and even created a coredump.  But unfortunately systemd-coredump limits size to at most 2GB, not enough for kstars:
BFD: warning: /home/pit/core.kstars.1000.9e88338e5e684b33aa6b617f427a7e42.16666.1625442302000000 is truncated: expected core file size >= 3329597440, found: 2147483648
I've increased the limit now - or are such backtraces not too helpful?

Other things I noticed:
  • EKOS scheduler used to print the twilight times for the current day, but no longer does 
  • The new colors of the additional catalogs have room for improvement.  E.g., I disabled the LDN catalog because that red color is completely killing the sky view (IMO, of course).  Unfortunately I found no place to change that color.  User error, or not implemented?
  • Right-clicking a target (e.g., to get the detail view), has always been a bit random when zoomed in.  But right now it's close to impossible for larger targets.  E.g., I view IC4628 at 4.9 degree FOV.  No matter where I click within that cyan diamond, it only tells 'Empty Sky'.  To 'find' the target, I first have to zoom out to around 37 degree FOV to get a 'match' on right-click on the second or third try....
  • For the above case (IC4628) the info-box in the upper right corner says (only) 'ESO 332-EN14' - not very helpful for most users, I'd assume.
The following user(s) said Thank You: Jasem Mutlaq
2 years 8 months ago #73226

Please Log in or Create an account to join the conversation.

The first issue was a fix by Eric a few days ago, so perhaps he can comment on it. The rest are still pending works by GSoC 2021 work by Valentin, hopefully he can take a look soon.

Regarding the crash, can you please run it via Ekos Debugger and share the backtrace for it?
The following user(s) said Thank You: Peter Sütterlin
2 years 8 months ago #73228

Please Log in or Create an account to join the conversation.

  • Posts: 1009
  • Thank you received: 133
Yeah, just wanted to report those somewhere - is there a specific issue tracker for his work, or is doing it here 'OK'?

As for the EKOS debugger - will have to see where to get it.  Is that on github somewhere?  Though a BT from gdb on the coredump (once I have a valid one....) should be OK, too, or not?
 
2 years 8 months ago #73234

Please Log in or Create an account to join the conversation.

  • Posts: 1009
  • Thank you received: 133
Hmm.   I even had it on disk already (ekosdebugger).  So I did a 'git pull' and generated a new executable.  But I don't get far.  It starts, I can select modules to track and launch kstars, and it starts logging some thread info, but then crashes
Cannot construct placeholder type QDBusRawType
Aborted (core dumped)
and the ekosdebugger window is gone.  gdb and kstars keep running, but as soon as I start up INDI, also kstars dies....
So I'll have to rely on coredump for now it looks.
 
2 years 8 months ago #73236

Please Log in or Create an account to join the conversation.

  • Posts: 1029
  • Thank you received: 301

Replied by Eric on topic Re:Issues with latest build

Concerning twilight, I changed the calculation of dawn and dusk: it was processed for the current day only and caused an issue when dusk happened before dawn.

Now dawn and dusk are calculated to be those next astronomical events following the startup time of the considered observation, shifted by the offsets configured in the Ekos options. If there is no astronomical dusk on the day of observation, the minimum altitude is used instead with a warning. Thus the values depend on the job that is selected in the table.

Now, if no job is selected in the table, my change was supposed to display the next astronomical dawn and dusk from the current Ekos simulation time (although de-selecting a line in the table is quite difficult). If you say something is amiss, I will double-check that point.

A minor point, if you schedule an observation for a night that has no dark time, Ekos should be able to push the job to the next proper interval, up to ten days from the current time.

-Eric
The following user(s) said Thank You: Peter Sütterlin
2 years 8 months ago #73239

Please Log in or Create an account to join the conversation.

  • Posts: 1009
  • Thank you received: 133
Thanks for the explanation, Eric!
What I was talking about was when you just open KStars, then EKOS, and go to the scheduler tab.  I.e., with no sequence etc. loaded.  For me this looks like this:

 

The old versions showed todays (rather: tonights....) twilight limits where the red scribble is.  I always check(ed) this when starting up to have an idea about available time.  Or is there another place in kstars to get this info easily?
2 years 8 months ago #73243
Attachments:

Please Log in or Create an account to join the conversation.

  • Posts: 1009
  • Thank you received: 133
No news on the crash.  Kstars ran without a single hickup all night.  Same binaries and sequence....
 
2 years 8 months ago #73263

Please Log in or Create an account to join the conversation.

I found and fixed the issue with Ekos Debugger, and it shoud work fine with the official KStars 3.5.4 release (pending today).
2 years 8 months ago #73265

Please Log in or Create an account to join the conversation.

  • Posts: 268
  • Thank you received: 72
Quoted from Peter Sütterlin post=73226 userid=4792
  • Right-clicking a target (e.g., to get the detail view), has always been a bit random when zoomed in.  But right now it's close to impossible for larger targets.  E.g., I view IC4628 at 4.9 degree FOV.  No matter where I click within that cyan diamond, it only tells 'Empty Sky'.  To 'find' the target, I first have to zoom out to around 37 degree FOV to get a 'match' on right-click on the second or third try
I updated to KStars 3.5.4. a couple of minutes ago and the mentioned behaviour above is still present. For certain objects (e.g. Andromeda) it's nearly impossible to hit the mark and invoke a slew with the right mouse button. This used to be my favorite and  fastest method to point the telescope to the desired object. Are there any enhancement in the line?
Last edit: 2 years 8 months ago by Toni Schriber.
2 years 8 months ago #73275

Please Log in or Create an account to join the conversation.

  • Posts: 268
  • Thank you received: 72
I found another small bug in 3.5.4: The device manager doesn't reflect the start of a driver! Subsequently it is not possible to stop it with the manager anymore.
2 years 8 months ago #73277

Please Log in or Create an account to join the conversation.

Thanks Toni, I actually traced it back to INDI and fixed it there. Thank you for the report!
The following user(s) said Thank You: Toni Schriber
2 years 8 months ago #73282

Please Log in or Create an account to join the conversation.

  • Posts: 268
  • Thank you received: 72
Thank you very much for your immense effort. KStars is an outstanding package!
BTW, what's the commit for this bug? I'd like to repair the manager, because I'm using it quit often for testing purposes.
2 years 8 months ago #73321

Please Log in or Create an account to join the conversation.

Time to create page: 0.667 seconds