Am 17.06.21 um 11:53 schrieb Nicolas George:
> Thilo Borgman (12021-06-11):
>> Am 11.06.21 um 21:07 schrieb Thilo Borgmann:
>>> Hi,
>>>
$subject seems to be useful depending on the network.
>>>
>>> v2 attached with updates to documentation and code.
>>> Please check if I missed the point of it
Thilo Borgman (12021-06-11):
> Am 11.06.21 um 21:07 schrieb Thilo Borgmann:
> > Hi,
> >
> > > $subject seems to be useful depending on the network.
> >
> > v2 attached with updates to documentation and code.
> > Please check if I missed the point of it.
>
> Follow-up to add the remark also to TC
Am 11.06.21 um 21:07 schrieb Thilo Borgmann:
Hi,
$subject seems to be useful depending on the network.
v2 attached with updates to documentation and code.
Please check if I missed the point of it.
Follow-up to add the remark also to TCP documentation.
-Thilo
From 69e6fb2acab42811c442eddafd
Hi,
$subject seems to be useful depending on the network.
v2 attached with updates to documentation and code.
Please check if I missed the point of it.
Thanks!
-Thilo
From 79f1589ba5d9224146d71c2605ca4cc02c852cb2 Mon Sep 17 00:00:00 2001
From: Nick Ruff
Date: Fri, 11 Jun 2021 21:02:02 +0200
Hi,
Am 09.06.21 um 12:41 schrieb Nicolas George:
> Thilo Borgman (12021-06-09):
>> $subject seems to be useful depending on the network.
>
> It looks premature. TCP_NODELAY should only ever used when the network
> code already takes care of writing data in a single system call. This is
> not what
Nicolas George (12021-06-09):
> Interesting thanks. I wonder why they did it like this, requiring two
> extra system calls for each flush; a flag on send would have served
> without extra system calls.
Ah, I see they have MSG_MORE, introduced at the same time, that does
just that. If it was portab
Steinar H. Gunderson (12021-06-09):
> If you want explicit flush, you can enable TCP_CORK (but it's Linux-only).
> Disable for the flush.
Interesting thanks. I wonder why they did it like this, requiring two
extra system calls for each flush; a flag on send would have served
without extra system c
Am 09.06.21 um 12:45 schrieb Steinar H. Gunderson:
> On Wed, Jun 09, 2021 at 12:41:17PM +0200, Nicolas George wrote:
>> (TCP_NODELAY is a terrible hack for a terrible API design. An explicit
>> flush system call would have been a much better choice.)
>
> If you want explicit flush, you can enable
On Wed, Jun 09, 2021 at 12:41:17PM +0200, Nicolas George wrote:
> (TCP_NODELAY is a terrible hack for a terrible API design. An explicit
> flush system call would have been a much better choice.)
If you want explicit flush, you can enable TCP_CORK (but it's Linux-only).
Disable for the flush.
/*
Thilo Borgman (12021-06-09):
> $subject seems to be useful depending on the network.
It looks premature. TCP_NODELAY should only ever used when the network
code already takes care of writing data in a single system call. This is
not what happens in ff_rtmp_packet_write() (libavformat/rtmppkt.c): t
Hi,
$subject seems to be useful depending on the network.
-Thilo
From 936826eec00ac2ceaa4579e8a6030a85191320a5 Mon Sep 17 00:00:00 2001
From: Nick Ruff
Date: Wed, 9 Jun 2021 12:26:03 +0200
Subject: [PATCH] lavf/rtmp: Add option to set TCP_NODELAY for rtmp
Suggested-By: ffm...@fb.com
---
doc/p
11 matches
Mail list logo
