The state of the memory-management community in 2024

Posted May 29, 2024 9:46 UTC (Wed) by bluca (subscriber, #118303)
In reply to: The state of the memory-management community in 2024 by taladar
Parent article: The state of the memory-management community in 2024

That's not what they tell customers. They tell them that their systems won't break by upgrading, while fixing known issues, which is the truth, and what users of long term running systems actually care about. That's very different from upstream kernel releases, which break compatibility all the time, and are extremely disruptive. It takes months of work to move from one upstream kernel version to the next for any real world use case and non-trivial workload.

The state of the memory-management community in 2024

Posted May 29, 2024 16:00 UTC (Wed) by tux3 (subscriber, #101245) [Link]

Hmm. For some less than all-encompassing definition of the real world.

I mean, that subset of customers who have to spend months fixing low level problems themselves? Not evenly distributed.
People delegate the dirty work to more competent dependencies. Rarely does anyone who lives on top of a big standard runtime notice anything when upgrading the kernel, with a very finite number of exceptions.

Those LTS customers who spend months firefighting whenever the seasons have the insolence to keep changing drown out the sound of everything else still humming along quietly.