Ok, I think I might know what's going wrong with the CSV import. Pixinsight is running on a windows host. AD runs from a Linux guest within VirtualBox. There is a shared filesystem between these two hosts (windows owns the disk). Linux maps a mount point to the windows drive. The windows side (Pixinsight) knows the directory as C:\Users\dsummers\Astro\Images\M3:

Here's one of the CSV lines showing the filename:
1,true,false,"C:/Users/dsummers/Astro/Images/M3/M_3_Light_001.fits",0.00000,2.76138,0.60201,0.77116,520.00000,56.07616,63.85677,0.97686,392,0.07981,0.27846,0.04923,0.08623

The linux side knows this same directory as:
/home/dsummers/AstroImages/M3 (note: AstroImages on linux vs Astro/Images on PC)

These are the same disk location. So, I think this must be part of the issue. I can import the files because the AD GUI internals likely map the mountpoint correctly, but is the hash or program somehow confused by mount point naming convention differences between the PC and Linux?
Hmm....what to do? If the only confusion is "Astro/Images" vs "AstroImages", I could correct this by changing the mount point in Linux to more closely match the PC. If it extends to "Users/dsummers", vs "home/dsummers", it's a bigger issue.

hi Doug,
design a resizable layouts using QT designer was a nightmare, at least to me. But I ended up with something usable. and resizable.
Precision was something easy to add even though Ekos exports coord data with just 3 digit precision, so it useless to increase precision on those data.
Moreover I will go for the sexagesimal / hour / deg visualization so precision is just needed in the database.

ferrante

AD tries to open (from your Linux VM) the file located at C:/Users/dsummers/Astro/...../.fits as written in the CSV file.
That's the only issue. Even if you mount with the exact same name I think it's the C:/Users that breaks up AD.

You could run AD from Windows? seems to be an easy solution.

Ok, I was able to confirm the CSV issue. A Windows CSV file must be converted to match Linux path convention. So, for example, each C:\Users\Image\blah line on PC must be converted to /home/Image/blah before AD will work correctly. Once I did this, AD was able to find the associated image file and import it (with correct results).

Not sure if you want to add a feature for this architecture nuance or not. Windows users can just push the CSV file to linux and run a script to convert, but it would be nice to have a "setting" for architecture specific CSV path conversion. If you do address it, I suggest two strings (from, to) as this will kill architecture AND any mount point naming differences.

Edit: Or as you say, run both from consistent architectures. I like the flexibility of a conversion string setting in AD, but now that we have mitigation(s), it's not a show-stopper.

Fantastic! Looking forward to using this.
Thanks for contributing.
Jo

hi Doug,
so you can see the charts now? good.
As you uploaded quite and amount of data, the app slowed down? I had up to 2000 images in the same db and it was ok.

I will release a new version tomorrow or the day after. Window resizing was tricky and also saving column width had some issues: saving columns was preventing the hide/show column to work properly.

Cheers
Ferrante

Uploaded on pypi a new version (0.2.7) that implements several suggestions mainly by Doug:
- Resize windows
- Save session for imagelist columns width and for window size and position
- Format RA/DEC, Alt/Az, Lat Long
- Precision (even though precision is affected by FITS header precision)
- Delete rows



Note about upgrading:
Please use a different virtual environment for testing if you previously installed AstroDom; or save your /astrodom/config files because installing in the same venv or using the upgrade.sh script will override some config files.
These are the config files affected:
- demo.db: no harm if you uploaded images to a different database
- config.json: upgrade will reset the active database selection
- profile-Ekos and profile-MaximDL: if you saved your own profile it will be ok.Else custom keywords will be overridden.


ferrante

Ferrante, I've updated to your latest. Thanks! The changes work as expected (I was able to save the changed field sizes, and resize the GUI). Much better. I am still getting a core dump on at least one specific use case. If you start to import a CSV file, but instead of following though, you "cancel" before loading the file, the app will core dump consistently. Here's the trace:

Traceback (most recent call last):
File "/home/dsummers/src/archive/AstroDom-master/resources/venv/lib/python3.8/site-packages/astrodom/importCsvTab.py", line 67, in importCsvFile
for row, val in enumerate(dataTemp):
UnboundLocalError: local variable 'dataTemp' referenced before assignment
./astrodom_linux_run.sh: line 18: 391996 Aborted (core dumped) python3 -m astrodom

I doubt this is a python3 thing.....probably want to chase that.

Cheers, Doug

Hi Doug, I noticed that error too. It s already fixed in 0.2.9 and pushed to github repo and pypi.
Thanks for your feedback.
Edit: it was not related to python3, I simply forgot to check for an empty selection when the file widget is closed .

So, before I do my own CSV path string conversion from Windows to Linux, are you sure you don't want AD to be architecture independent for this particular detail? Grabbing the path string from the CSV and converting to local form for all might be better than me doing this for myself only... As I said before, the preferred method would be two strings (to,from) to accommodate filesystem mount point differences.

A second question for you is whether the astrodom_linux_upgrade.sh script down in resources works correctly for minor updates. I did the full pull/install last time, and if the update script works more simply for minor patches like this latest one, that would be preferred. Can you comment on that script? Thanks.

The upgrade script works the same way as install except the first doesn't recreate the 'venv' virtual environment. So yes, it is more suited for small updates . Just be aware that the config files will be overwritten in both cases, so prepare a back up ( I will fix this soon because it's annoying ). The database should not be affected but I would do a backup in any case.

Ok I will implement the path conversion because it happens quite often to move files around. It will be set in the settings tab and in the csv tab you can check/uncheck if conversion has to be applied.

now with v 0.2.11 you can replace (part) of a path to search for CSV file.
Path input is in CSV import tab. Was a better location than Settings.

Usage example:
Actual FITS file is on Windows: C:\users\ferrante\IC434\Luminance\test.fits
CSV file was made when on Linux: /home/ferrante/IC434/Luminance/test.fits
So replace From: /home/ferrante/
To: C:\users\ferrante\

- It works the other way around too (FITS saved on Linux, CSV on Windows) and of course with both coming from the same OS.
- Don't worry about creating mixed path like C:\users\ferrante\IC434/Luminance/test.fits AstroDom will fix it.
- It relies on a function that detects the OS separator. Only tested on Linux, hope it works on Windows too.
- If From is not empty while To is, replace is still applied with an empty string. If From is empty no replacement is applied.