Avoid flushing headers when the server returns a single response#9314
Avoid flushing headers when the server returns a single response#9314sanjaypujare merged 4 commits intogrpc:masterfrom
Conversation
ejona86
left a comment
There was a problem hiding this comment.
binder/ also needs updating. You didn't notice it before because -PskipAndroid=true will skip it.
|
The history has gotten convoluted here and seems broken, and is breaking the diff. Are you fine with me cleaning it up? (Doing so will require force-pushing to the branch, and I'm not entirely sure if I have access to do that but I can certainly try.) |
ejona86
left a comment
There was a problem hiding this comment.
I went ahead and cleaned up the history to fix the diff.
|
@amirhadadi Would you please complete the EasyCLA and rebase to resolve the conflicts? |
|
@larry-safran as we traced the issue of extra segments being sent to flow control PINGs, not sure it's worth merging this PR. |
|
With #9847 , we won't be sending PINGs as frequently, especially for small requests/responses on a warmed connection. Seems this wasn't merged just waiting on @sanjaypujare's review. |
Yes, I noticed that a couple of days ago. Is there still interest in merging this PR? From yours and @amirhadadi 's comments I am guessing this PR may not be needed anymore. Let me know |
|
My comment was to say this has use, as the thing that made it less useful has been fixed. (And even back when it didn't demonstrate a win, I was approving of it.) |
|
Okay. @amirhadadi do you want to rebase then I will review. 3 files have conflicts |
…rver has a single response.
resolves #4884