To celebrate #38C3 I finally worked on my backlog and fixed the #Threema #Flatpak that I... ugh... procrastinated for 9 months.
It took rather long because curiously the nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpqdxqs0nqka84wfh357p8knlayx4u74t5692zftlp8tunjmfyf7sfspj4jse (nprofile…4jse) CI wouldn't build the package because of EAI_AGAIN which usually indicates that there's a build step requiring Internet when the CI denies it. In theory, flatpak-node-generator takes care of this by pre-downloading the packages out-of-band and indeed, the actual build runs perfectly fine. However, one step in packaging the Electron app as ASAR, for some reason, wants to connect to GitHub.
The thing is: This error was there before and did not crash the CI. And indeed, on every 3rd or 4th try, the CI would just... continue despite this build step throwing an error 0:-)
Yes, there's a race in the build script + CI tooling that allowed the CI to continue building the Flatpak sometimes, despite the build step error. And ontop of this: This was precisely the build step we actually do not need (we need everything /until/ the final ASAR bundle is created but not the bundle itself). So I just removed the `exit(1)` from the exception handler in the build script and carry on...:
https://github.com/flathub/ch.threema.threema-web-desktop/pull/13/commits/c753be0872790c7bb4702772e7a585400ca9eb3d
Sorry nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpqzg9npae5xc538uvul2weljxgs6gyzhuecghw3v42jfxd3r3jyf2sg4ry7v (nprofile…ry7v), I really don't want to debug electron-packager + Flathub CI 0:-)