Alex Rousskov
Alex Rousskov
> The issue I am trying to fix is that the AcceptLimiter::kick() call to TcpAcceptor::acceptNext() does not have AsyncCall protections for proper exception handling Understood. Thank you for disclosing this....
> One other implementation would be to have each TcpAcceptor maintain its own deferred queue instead of a global AcceptLimiter instance. > You see any flaws with that design? I...
> Alex: The best solution I can suggest replaces the explicit while loop with an implicit one, done via a special TcpAcceptor callback data member. I found a much simpler...
> I did not review the code changes. They could use some style polishing, but I hope others will take care of that. @Opendium, if this does not happen in...
BTW, you do not need to squash and force-push your changes during review. [Anubis](https://github.com/measurement-factory/anubis#readme) will squash your changes automatically when it is time to merge your branch into the official...
@yadij, I could not find the answer to this earlier review question so I am repeating it here for your convenience. AFAICT, SslBump still requires a signing certificate for nearly...
Your answer discusses internal implementation details, but I was asking about the configuration interface. Since we seem to have trouble communicating, I will try to split my question into several...
>I am waiting for a good conclusion to the thread from [#827 (comment)](https://github.com/squid-cache/squid/pull/827#discussion_r638896499). That will determine whether the other open change requests happen. If you are waiting for me, my...
> This PR is blocked by @rouskkov self-requested review and "I will do my best not to participate in further discussions" decision. Both assumptions are false and the misleading quote...