Changes by Andrej A Antonov <polymor...@gmail.com>:
--
nosy: +polymorphm
___
Python tracker <rep...@bugs.python.org>
<http://bugs.python.org/issue24755>
___
Andrej A Antonov added the comment:
I just will write next code-fragment:
import socket
s = socket.socket(socket.AF_INET, socket.SOCK_STREAM, socket.IPPROTO_TCP)
s.connect(('python.org', 80))
print(
'is my operation system using (by default) tcpkeepalive-algorithm
Andrej A Antonov added the comment:
in GNU/Linux system timeout has been reached -- means that system timeout
will *never* reached.
That's quite likely because the system limits may be very large.
I tested system-timeout GNU/Linux (on various computers). I waited more then 5
days. system
Andrej A Antonov added the comment:
@demian.brecht , for high probably to catch *infinite_freeze* (at GNU/Linux) --
if we may will run requests of xmlrpc.client.ServerProxy -- parallely:
(when running next code -- need to make some network-disconnections on
network-router-computer
Andrej A Antonov added the comment:
good patch (issue14134.patch) ! thanks!
--
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue14134
___
___
Python
Andrej A Antonov added the comment:
@demian.brecht , socket.setdefaulttimeout([timeout]) -- it is bad practice,
because setting this global varible we may spoil other cases. example TCP
keepalive [ s.setsockopt(socket.SOL_SOCKET, socket.SO_KEEPALIVE, true) ]
and global variables is bad
Andrej A Antonov added the comment:
ok, let's go to other side of this problem:
question: why default transport (xmlrpc.client.Transport()) is not setting
value of timeout?``
answer: because *unknown* which value need to using by default.
in various cases programmer need various timeout
Andrej A Antonov added the comment:
if you NOT point timeout in RPC-client -- you program will freeze or will
maked resource leak (with small probability).
Assuming a lack of concurrency, your program will indeed freeze until the
system timeout has been reached. I'm not sure about a leak
Andrej A Antonov added the comment:
@demian.brecht , your example code-fragment is too big. :-)
too many lines -- just only for adding timeout. it is uncomfortably.
most people will not using that: most likely they just will forget about
timeout (but in *MOST* situations not using timeout
Changes by Andrej A Antonov polymor...@gmail.com:
--
nosy: +polymorphm
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue21356
___
___
Python-bugs
Andrej A Antonov polymor...@gmail.com added the comment:
Jeff McNeil (mcjeff) I would think it might make more sense just to make the
change to the Transport object. Since there's an argument for a transport on
ServerProxy already, that seems more straightforward and keeps the network
layer
Andrej A Antonov polymor...@gmail.com added the comment:
in this subject -- I think about like this changes (see file
example-of-changes.patch)
--
keywords: +patch
Added file: http://bugs.python.org/file24675/example-of-changes.patch
___
Python
New submission from Andrej A Antonov polymor...@gmail.com:
good day!
xmlrpc.client.ServerProxy() -- not has timeout-parameter
xmlrpc.client.Transport() and xmlrpc.client.SafeTransport() -- not has
timeout-parameter too
but http.client.HTTPConnection() and http.client.HTTPSConnection() -- has
13 matches
Mail list logo