Changes between Initial Version and Version 1 of Ticket #22212, comment 8


Ignore:
Timestamp:
May 26, 2017, 1:44:03 AM (18 months ago)
Author:
mikeperry
Comment:

Legend:

Unmodified
Added
Removed
Modified
  • Ticket #22212, comment 8

    initial v1  
    33I am not sure what to do about this. I could change the code to always schedule an actual timer right away and cancel it if data is sent instead (instead of waiting around until we get within a second before scheduling a timer). This may mean more timer churn, but now that the timers are more efficient, perhaps this is not an issue.
    44
    5 Or, we could demote the log message to notice.
     5Or, we could demote the log message to notice.
     6
     7In terms of when padding actually gets sent, I'm guessing that it still won't happen any earlier with the schedule and cancel timer approach, because if this is #16585, libevent wouldn't call us back any sooner anyway.
     8
     9This is making me think that we just demote this message and not increase the code complexity.. Any objections?