Yes it does. Thank you very much.
Hey everyone, I’m trying to get a new version out soon, I know it’s been far too long but this version that will migrate peoples data has me shaking in my boots even if my code is probably fine.
Since I am so worried I’m thinking I will stagger the release more than normal in the sense that I will first upload to OpenRepos and wait for 1-2 weeks before submitting to Jolla Store QA.
I would also add some warning test about beta testing on OpenRepos.
WDYT?
13 Likes
You seem to have prepared both with dry-run in current version and manual lever-pulling for the actual migration, so why don’t we just covert a number of databases, with db backup copy so the conversion can be run again if needed, e.g. fixes for some initial hiccups if any? Whatever warnings you put there people will just ignore, so make it easy to do the right thing, and keep the initial amount of feedback managable but diverse enough to get some fixes in. It is a delicate task, but without data from at least all dimensions of db inconsistency it is hard to get the relaxed feeling to show up.
What do you mean?
It’s changing the path at which the data lives nothing else (the db stays as is), I guess I could make a package available for download to those who request it in a DM and see how that goes.
You tell me. I thought the conversion would take as long as the currently displayed migration checks, which is usually 10-20 minutes in my case, but it might be the core starting independently of the migrations taking the most time. It sounds you have it covered then. A good start. So please DM whenever you feel there is something to test.
10-20 minutes??? 
The checks just stay on the screen while podcasts are loading, on my phone (10ii) with 34 podcasts and ~4600 episodes it takes all in all ~10 seconds to get to the main screen.
How large is your library?
/Edit -
Note that the last line of the output that you see is what is actually happening at that time, most likely it is “Loading podcasts”
Hehe, yes, I have been enjoying gPodder for a long time. I did restart 4.18.0 on my C2 when I wrote my previous reply and it was at 1 hour on “Starting kernel” (or similar in english) as the message at the bottom, before I killed it. I will check number of episodes, but I removed thousands to get the loading time more reasonable, it is 84 GB in the data folder (downloaded episodes seems to be around 2200). On a new attempt I actually got “Loading podcasts” quite quickly, and now 10 minutes later it finally showed the main screen.
1 Like
Wow that is impressive, most of my episodes are not downloaded.
During startup gPodder is scanning the folders so that would slow things down, there is probably room for improvement there though that is not going to be in this release.
Is all that data on your phone or on an SD card?
1 Like
I did move it to a SD card after the spring-cleaning. It was not quicker on phone storage. Here are the stats from the cover: # podcasts: 97, episodes: 20642, new episodes: 149, downloaded: 1358
Hi everyone,
So I thought I had this whole migration thing nailed… everything worked nicely in the SDK and then I ran it on my phone and I discovered the SDK runs programs differently than the phone.
While I think I have mostly ironed out the kinks now I also discovered something that I really don’t like -
Due to the way the sandbox works when migrating from inside the sandbox everything gets copied instead of simply renamed this means that if you have a large collection you could run into space issues and regardless of the size you have a potentially large amount of unneeded writes to your phones storage happening shortening its’ life.
Unless there is a “Jolla approved” solution to this I am inclined to publishing a simple bash script to move your folders manually so that those who read release notes and value their hardware longevity can do this while the rest will go through a X minute migration…
Would like to hear what you think.
Also my current journey is tracked here - Understanding SailJail flow - #11 by Keeper-of-the-Keys
4 Likes
Actually I don’t need to write a separate script, just recommend people manually start gPodder outside of the sandbox once for migration.
That would basically require the user to start gpodder from the terminal with sailfish-qml harbour-org.gpodder.sailfish
I’ll also work on better feedback that something is happening as things will be slow for those with large collections.
Starting gPodder once without sandbox is doable for me?
- Yes
- No
0
voters
1 Like
As english is not my native language, this discussion thread is not easy to understand every thing. Above all, i’m not a developper.
Anyway, i love gPodder 
At this time, all my datas are on my µSD card (~30 Go of datas !). And gPodder find only the ~15 last days. But I did select the download folder in the interface.
So, i would likeif this kind of workaround is always ok today
? Or not ?
In version 4.16.2 I added support for configuring where you store your downloads so that you don’t need to do that trick (unless you also want the database to be stored on your uSD card).
While I suspect the trick would continue to work I can’t promise it and due to the way the migration logic works it is very likely that it would end up moving everything back from the SD card to your phone if you don’t manually migrate by yourself.
3 Likes
Thanks a lot for your answer.
I see that on /home/deaultuser/.local/harbour-org.gpodder.sailfish i have two files:
.directory 33o
Database.minidb 5.67mio
And in /run/media/defaultuser/0ca….1
/harbour-org.gpodder.sailfish i have three files
.directory 33o
Database.jsondb.migrated : 76.84Kio
Database.minidb: 3.50Mio
I think that there is a “conflict” between the two Database.minidb.
Do you think that il would erase the Database.minidb in /home/deaultuser/.local/harbour-org.gpodder.sailfish ?
Thanks for your advise.
I’m very confused, didn’t you say you used the symlink method you linked to?
With a symlink the contents of both paths would be identical.
The minidb files are just sqlite files so you could copy them to your computer and use a sqlite browser to determine which one of them is the database you are using.
Hi, gPodder is a great audiocast app. I love you!
But I would still have a wish: Is it possible that you could start gPodder in the MOST NEW mode and define this as a home page?
I also wish I could update the LATEST page via a pull-down menu. Is that possible?
Best regards
I’d’ve answered yes but the poll is closed already.
Though I’d still prefer a bash script, even if it does little more than starting gPodder once without sailjail, plus some checks.
if [[ "$migrated" == true ]]; then exit 0; else .....
Why is a bash script better than just starting the application in your view?
From my end it doesn’t look so different just trying to understand other viewpoints.
I was thinking about this after I wrote it and honestly, it was bs.
I was thinking it would be more idiot proof, but if your method is documented sufficiently it makes no difference.