https://bugs.exim.org/show_bug.cgi?id=400
Phil Pennock <pdp@???> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |pdp@???
--- Comment #5 from Phil Pennock <pdp@???> ---
I am hesitant to see shutdown be the default, as it's a change in behavior in
an area with known compatibility issues for some services. It's a very neat
and clean solution, but one which breaks some widespread assumptions.
Eg, nginx considers a client-side shutdown to be a "client abort" and is not
alone in doing so; for the SKS project, I had to document using
"proxy_ignore_client_abort on;" for SKS keyservers, to deal with older versions
of the GnuPG keyserver client and it's mock-curl library.
These problems are hard to debug.
So: a change in default which would be hard to debug and lead to
timing-sensitive incomplete data? Using a TCP feature which ~90% [WAG] of
network programmers don't know about? That seems bad.
As an option which can be explicitly set, and read as "send this data and then
shutdown", that's much easier to see and reason about and know it's happening.
--
You are receiving this mail because:
You are the QA Contact for the bug.
