Hi, On 11-01-2025 10:47, Niels Thykier wrote:
... if I had still been a member of the RT, ...
Thanks for raising this issue, which no doubt was inspired by your work in the team. I think you gave a great description here of the balancing act we need to perform in our team. Obviously the limit lies somewhere and thinking about where that is might be good. It's my feeling too that the ballpark limit for yo-average-package should be around 48 hours.
Given that the porters expect the situation for riscv64 to improve in the not so distant future and that riscv64 *currently* is much slower than the other architectures, I'm not sure I want to dive deeper into where exactly to draw the line, as I'm not going to codify it (soon) and without code to check it, I'm not sure we already need to detail everything out. We might not need to check on this in a while.
So, my view on this is: yes in general (exceptions possible) an architecture should be able to build binaries within a reasonable time. 48 hours seems a reasonable time at this moment to me.
Paul
OpenPGP_signature.asc
Description: OpenPGP digital signature
