Request 1070697 (accepted)

Overview

Request 1070697 accepted

No description set

Created by david.anes about 1 year ago
In state accepted

Submit libdb-4_8

Comments for request 1070697 5

Jan Engelhardt (jengelh) - about 1 year ago

reviewer source maintainer

Is this really useful? DB is more IO-bound than compute I would assume. Not only that, but this is libdb 4.x, something that's ancient by now.

David Anes (david.anes) - about 1 year ago

author source maintainer

We are reviewing and enabling in all packages that we can that could benefit from this. I understand @pluskalm did review and found this may be beneficial. In any case, it won't hurt.

Fabian Vogt (favogt) - about 1 year ago

In any case, it won't hurt.

Oh, it can. Slowdowns are very much possible.

David Anes (david.anes) - about 1 year ago

author source maintainer

It's hard for me to find a scenario where that can hold true in this specific scenario:

If it's certainly io-bound, then the (possible) slowness will be orders of magnitude less than perceptible or even measurable.
If it's cpu-bound, I understand then it will be beneficial.

In any case, maybe @pluskalm can add more info here, as he did the request?

Jan Engelhardt (jengelh) - about 1 year ago

reviewer source maintainer

It “hurts” in a different place. More variations, more building, more packages, more metadata (repomd.xml) to process by zypper.

https://build.opensuse.org/request/show/1071411 even admitted no measurement, so now everything is under double scrutiny.