Increase bed temperature to 100C, didn't help much - the bed did get appreciably warm, but the test cube still had two corners lifted:
Trying now with an UHU stic [sic] on warm bed. That stuck like crazy - in fact the cube stuck so much I scratched the bed badly trying to get it off. In the end I put the plate in the freezer, after which I could get the cube off. But taking the bed to the freezer after each print is a bother and waste of energy.
With the chilled bed, I tried with hairspray and gluestick both. No sticking with either. However, that does indicate that with a temperature in between cold and hot, gluestick should work.
Tried with setting the bed temp to 30C for a while, until it was lukewarm. That gave a removable but still properly cornered piece.
Unfortunately, the bed starts off colder than this, otherwise it would have been nice to be able to not do any bed heating at all.
I'll probably get a glass plate, as it has a wider range of good adherence using hairspray, but I'll keep the aluminium for things that have small footprints. For instance, this replacement floater from one of my plant box:
To try something flatter, here are some mounts for the tomato rail. They were a bit difficult to get off, they stuck enough to bend a bit.
The need for glue stick is annoying and hard to get right. There's not a lot of leeway between too stuck and curled corners. So I went to the local glazier and got a new piece.
The new bolt had different enough dimensions that I ended up flipping the big gear and just use tension to keep it in, rather than having a nut on the outside in the nut trap. This is probably why it stripped fairly quickly. It stripped with a lot more powdered filament than the old bolt. Now I'm using the nut with fastening screw on the far side and a nylock on the near side. Would have been nice if the bolt itself had flats at the end for fastening nuts.
Finally just found the right place for a nut and epoxied it on.
-- Vacation break --
With the now well-hardened nut on the bolt, it's easy to put the whole thing together right, and extrusion looks good. It did another of the really slow Z moves, I still don't understand why that happens. When homing, it went at full tilt.
Z offset 0.6. Printing simple cube without any hairspray, because this glass plate seems extra smooth, and I want to see if the idea that the adhesion is better at high smoothness has any merit. Not really, the corners are still turned up a bit.