I didn’t mean comparing the contents, just whether you delete all .trsp files or just the ones that have data in transmitters.json?
My concern is if the user adds his/her own .trsp file for a satellite that does not have data in Satnogs DB, will this procedure then delete this trsp file or leave it in place?
Just take your time because I’m busy the next two weeks anyway
Regarding your question to @pierros about next steps, I’m hoping we can merge / cherry pick this functionality into my main repository so that it becomes part of Gpredict.
Today I have merged the changes done by Baris @ta7w along with some fixes of my own (mostly of practical and cosmetic nature). Good work Baris and thanks a lot.
I have also disabled including of the .trsp files in the gpredict packages, so the primary frequency data source is now Satnogs DB. The files are still in the git repository for reference, in case we find some sats are missing from the DB, but they are no longer distributed with gpredict.
I had to revert the changes because we found a few issues that make it not quite ready for prime time…
The JSON file lists transponders one at a time and so the import code will have to merge different transponders for the same satellite. Currently, only the last transponder makes it to the .trsp file.
The mode in the JSON file is an “id” and not a human readable string. I wonder if this is an error at the database end, i.e. the code that generates the JSON file.
I wonder if it would make sense to group transponders together per satellite in the JSON file?
All changes that were reverted are now in a separate satnogsdb branch so we can continue working on it there.
The “mode_id” is a key to a different table/endpoint accessible here: SatNOGS DB API (for sanity of data reasons).
@ta7w I am wondering if you can concat different transceivers with common norad_id in the same file, while resolving the mode_id to human readable format using SatNOGS DB API ?
That could be used too after the list of satellites was fetched.
Anyway, I don’t think it is a problem. It just means that the function in gpredict will have to do a little more than traversing the list once and writing data as it encounters it.
Yes, there is an open pull request now that is waiting for me, but I am overloaded with gqrx at the moment and probably till the end of September, sorry.