Ferenc Erki

> Did you two fetch that database? (Or the other way around: Are you still interested in keeping that download available?) @jbglaw: I got a copy, thank you!

Using the downloaded copy, I don't seem to be able to (fully) reproduce the issue. I feel like I might miss some crucial detail in how to reproduce, or my...

> With that setup (using a NVMe disk), you can probably read the whole file within a few seconds. I've got way more RAM, but that's actually also in use....

> Even if the log's in a separate file, the browser is going to struggle fetching and ansi-highlighting a ~900Mb log. Yes, I guess that's what breaks my job output...

I took a closer look at SQLite internals, and I feel I can clarify some of our findings so far. ## SQLite internals SQLite, like many other databases, stores data...

After letting the topic rest for a while, I think there are at least two more aspects that I was not able to investigate fully yet. ## Column order The...

> ## Column order > The patched database schema has the BLOB `output` as last column in the rows, and maybe that's why it's able to skip read the overflow...

Thanks for this cleanups too, @fkobi! For the record, as proxied maintainer of media-plugins/alsaequal, I agree with the proposed changes in that ebuild (bump copyright year, bump EAPI version, remove...

> 1. It is almost completely neutralized with `unsafe-eval` and `unsafe-inline` That's right, and I believe a part of the story is about improving the situation for users with high...

> Keep in mind that if you actually want to restart the laminar process (you mentioned "a new version of Laminar or something") then jobs in the queue are lost...