Re: [Capy Review] Why no default rescheduling?
Steve Gerbino via Boost <[email protected]> Fri, 26 Jun 2026 21:35:25 +0000
| Newsgroups | gmane.comp.lib.boost.devel |
|---|---|
| Message-ID | <ZsxuZtZ97pbg8hNaPH0CNVEQ2CkiS69LIQCwpGi5Rh7HXn6V63YHawC4zX0y-Artltx5rR2egbBkOZOuZ5GdtWxtSG9G9mqloyONpaJ2csw=@gerbino.co> |
On Friday, June 26th, 2026 at 11:17 PM, Marcelo Zimbres Silva via Boost <[email protected]> wrote: > I would like to know why Corosio/Capy authors decided to take this > direction? Symmetric transfer was preferred for two reasons. 1. Symmetric transfer ensures that we don't have unbounded stack growth 2. It is more performant The networking scenario you described in which an adversarial connection monopolizes the event loop (which would fall under the purview of Corosio) is mitigated through a mechanism we call the inline_budget. The inline_budget is how many consecutive symmetric transfers are permitted before we should fall back to posting in order to ensure the scheduler does not get starved. _______________________________________________ Boost mailing list -- [email protected] To unsubscribe send an email to [email protected] https://lists.boost.org/mailman3/lists/boost.lists.boost.org/ Archived at: https://lists.boost.org/archives/list/[email protected]/message/FCZRFJCX2ZSN4DDINVCJH3GMWHIH2CDI/