Attention is currently required from: neels, pespin.

View Change

1 comment:

• Commit Message:

  • Patch Set #3, Line 12: Use OTC_SELECT

    I personally suggest using stack-allocated buffers for those cases where we don't have/want static buffers.

    Agreed.

    But I'm really surprised to hear any praise for static buffers at all.
    I thought that the consensus and aim is to get away from static buffers

    I'm not aware of such a notion/consensus. The heap allocated variants to me only make sense in cases where they don't work for some specific reason.

    Also, it is a big difference whether some function is called once (like to generate a fsm name, or some other name of an object that has a certain lifetime) in a while, or a function is called potentially thousands of times per second.

    IIRC there is a log line that prints two distinct cell IDs

    If you know that, then I would suggest to special-case that one statement.

    If you mean to imply that this is inefficient and expensive related to the rest of osmo-hnbgw

    I don't want to make such decisions specific to each and every program. The fact that there may be some other horribly inefficient parts in osmo-hnbgw shouldn't be seen as an excuse to introduce more *unless* there's a clear advantage.

    Also, given that the user triggering this development is very keen on efficiency, and we are currently spending an awful lot of time on osmo-bsc optimizations. Sooner or later I expect the same will happen with osmo-hnbgw, as it gets loaded.

To view, visit change 33169. To unsubscribe, or for help writing mail filters, visit settings.