Re: gethostbyname_ex(hostname) extremely slow (crossposted from stackoverflow)
On Monday, September 2, 2013 5:45:26 AM UTC-7, Roy Smith wrote: In article 00843d58-db21-4cf0-9430-85362a1dd...@googlegroups.com, anntzer@gmail.com wrote: As it happens I found a better way: just add the proper entry to /etc/hosts. You have not found a better way. You still have a network (or more specifically, DNS) configuration that's broken. What you have found is a pragmatic way to solve your immediate problem and get some work done. That is certainly useful (and I've done it plenty of times), but you need to understand that what you've done is hidden the problem, not solved it. To be honest, knowing nothing about DNS configuration, I don't even know if adding the entry to /etc/hosts is the proper fix or if the issue should be fixed somewhere else (or perhaps didn't know, as you seem to imply that that is not the correct way). -- http://mail.python.org/mailman/listinfo/python-list
Re: gethostbyname_ex(hostname) extremely slow (crossposted from stackoverflow)
On Saturday, August 31, 2013 10:06:43 PM UTC-7, Michael Torrie wrote: On 08/31/2013 10:51 PM, anntzer@gmail.com wrote: It is the call to gethostbyname_ex that is very slow. The call to gethostname is quick (and returns the same string as /usr/bin/hostname). What gethostbyname_ex and /usr/bin/hostname do are very different things. gethostbyname_ex does a DNS lookup against a server. /usr/bin/hostname just checks a local computer setting. I don't see why you are comparing the two. /usr/bin/hostname is not going to help you find a list of IP addresses that point to a machine. I was just replying to the previous comment name = socket.gethostname() see how long that takes and what it returns. Then, assuming it returns a string containing your hostname (massive handwave about what that actually means), saying that gethostname resolves my own hostname instantaneously. -- http://mail.python.org/mailman/listinfo/python-list
Re: gethostbyname_ex(hostname) extremely slow (crossposted from stackoverflow)
On Sunday, September 1, 2013 4:37:34 AM UTC-7, Chris Angelico wrote: Yes, it most definitely CAN be a network config issue. The C function you want to be calling is getifaddrs(), and I don't think there's a way to call that from core Python. But a Google search for 'python getifaddrs' shows up a few third-party modules that might be of use to you; that'd be a lot quicker and more reliable than trying to look up your own hostname and depending on the results. ChrisA I tried using netifaces (https://pypi.python.org/pypi/netifaces) which seems to rely on getifaddrs (according to the doc, I didn't check the source). Again, it returns nearly instantaneously the correct IP address. -- http://mail.python.org/mailman/listinfo/python-list
Re: gethostbyname_ex(hostname) extremely slow (crossposted from stackoverflow)
On Sunday, September 1, 2013 2:03:56 PM UTC-7, Chris Angelico wrote: I tried using netifaces (https://pypi.python.org/pypi/netifaces) which seems to rely on getifaddrs (according to the doc, I didn't check the source). Again, it returns nearly instantaneously the correct IP address. Perfect! ChrisA Not really for my use case -- it isn't that *I* want to know my public IP address, but rather that IPython wants to know it. Of course I could patch IPython's source to use netifaces but that sounds like an overkill. As it happens I found a better way: just add the proper entry to /etc/hosts. Still, thanks for the suggestions. Antony -- http://mail.python.org/mailman/listinfo/python-list
gethostbyname_ex(hostname) extremely slow (crossposted from stackoverflow)
Hi, At startup, IPython (qtconsole) calls socket.gethostbyname_ex(socket.gethostname())[2] to find a list of IP addresses that point to the machine. On a Linux server that I manage this call is extremely slow (20s)... which I have trouble understanding as ip addr show seems to give the same information nearly instantaneously. Is there anything I can do to make this faster? Can this be a network configuration issue (I am behind a router)? This issue is independent of IPython: $ time python -c 'import socket; print(socket.gethostbyname_ex(socket.gethostname())[2])' ['192.168.0.102'] python -c 0.07s user 0.02s system 0% cpu 28.190 total Thanks. Antony -- http://mail.python.org/mailman/listinfo/python-list
Re: gethostbyname_ex(hostname) extremely slow (crossposted from stackoverflow)
It is the call to gethostbyname_ex that is very slow. The call to gethostname is quick (and returns the same string as /usr/bin/hostname). On Saturday, August 31, 2013 6:01:00 PM UTC-7, Roy Smith wrote: In article b9f77b6f-3a65-407a-aff5-5677be2ba...@googlegroups.com, anntzer@gmail.com wrote: Hi, At startup, IPython (qtconsole) calls socket.gethostbyname_ex(socket.gethostname())[2] to find a list of IP addresses that point to the machine. On a Linux server that I manage this call is extremely slow (20s)... which I have trouble understanding as ip addr show seems to give the same information nearly instantaneously. Is there anything I can do to make this faster? Can this be a network configuration issue (I am behind a router)? It's almost certainly some sort of configuration issue which is causing some DNS query to timeout. The first step, is to figure out which part is taking so long. Open up a python shell and run: name = socket.gethostname() see how long that takes and what it returns. Then, assuming it returns a string containing your hostname (massive handwave about what that actually means), try socket.gethostbyname_ex(name) and see how long that takes and what it returns. At least at that point you'll have cut the problem in half. If I had to guess, you've got a missing PTR record, because that's what most people screw up. -- http://mail.python.org/mailman/listinfo/python-list
issubclass(C, Mapping) not behaving as expected
from collections import * class C(object): def __iter__(self): pass def __contains__(self, i): pass def __len__(self): pass def __getitem__(self, i): pass issubclass(C, Mapping) = False [issubclass(C, cls) for cls in Mapping.__mro__] = [False, True, True, True, True] i.e. C does implement Sized, Iterable and Container. I would have expected that just as issubclass(C, Sized) checks for the presence of a __len__ method, issubclass(C, Mapping) would check for the presence of the three methods required by each immediate superclass? Antony -- http://mail.python.org/mailman/listinfo/python-list
Re: cmd2, an extenstion of cmd that parses its argument list
On Sunday, March 18, 2012 10:12:24 PM UTC-7, anntz...@gmail.com wrote: Dear all, I would like to announce the first public release of cmd2, an extension of the standard library's cmd with argument parsing, here: https://github.com/anntzer/cmd2. Due to an already existing Cmd2 on PyPI, I have renamed the project to parsedcmd, which is also a better description of what the module does. https://github.com/anntzer/parsedcmd Cmd2 is an extension built around the excellent cmd module of the standard library. Cmd allows one to build simple custom shells using ``do_*`` methods, taking care in particular of the REPL loop and the interactive help. However, no facility is given for parsing the argument line (do_* methods are passed the rest of the line as a single string argument). With Cmd2, ``do_*`` methods are type-annotated, either using Python 3's function annotation syntax, or with an ad-hoc ``annotate`` decorator, allowing the dispatcher to parse the argument list for them. Antony Lee -- http://mail.python.org/mailman/listinfo/python-announce-list Support the Python Software Foundation: http://www.python.org/psf/donations/
Re: ANN: cmd2, an extenstion of cmd that parses its argument line
I have renamed the project to parsedcmd, which is also a better description of what the module does. https://github.com/anntzer/parsedcmd On Monday, March 19, 2012 6:14:44 AM UTC-7, xDog Walker wrote: On Sunday 2012 March 18 22:11, anntzer@gmail.com wrote: I would like to announce the first public release of cmd2, an extension of the standard library's cmd with argument parsing, here: https://github.com/anntzer/cmd2. There already is a cmd2 package at PyPI and has been for a long time. http://packages.python.org/cmd2/ -- I have seen the future and I am not in it. -- http://mail.python.org/mailman/listinfo/python-list
Re: cmd2, an extenstion of cmd that parses its argument list
On Sunday, March 18, 2012 10:12:24 PM UTC-7, anntz...@gmail.com wrote: Dear all, I would like to announce the first public release of cmd2, an extension of the standard library's cmd with argument parsing, here: https://github.com/anntzer/cmd2. Due to an already existing Cmd2 on PyPI, I have renamed the project to parsedcmd, which is also a better description of what the module does. https://github.com/anntzer/parsedcmd Cmd2 is an extension built around the excellent cmd module of the standard library. Cmd allows one to build simple custom shells using ``do_*`` methods, taking care in particular of the REPL loop and the interactive help. However, no facility is given for parsing the argument line (do_* methods are passed the rest of the line as a single string argument). With Cmd2, ``do_*`` methods are type-annotated, either using Python 3's function annotation syntax, or with an ad-hoc ``annotate`` decorator, allowing the dispatcher to parse the argument list for them. Antony Lee -- http://mail.python.org/mailman/listinfo/python-list
cmd2, an extenstion of cmd that parses its argument list
Dear all, I would like to announce the first public release of cmd2, an extension of the standard library's cmd with argument parsing, here: https://github.com/anntzer/cmd2. Cmd2 is an extension built around the excellent cmd module of the standard library. Cmd allows one to build simple custom shells using ``do_*`` methods, taking care in particular of the REPL loop and the interactive help. However, no facility is given for parsing the argument line (do_* methods are passed the rest of the line as a single string argument). With Cmd2, ``do_*`` methods are type-annotated, either using Python 3's function annotation syntax, or with an ad-hoc ``annotate`` decorator, allowing the dispatcher to parse the argument list for them. Antony Lee -- http://mail.python.org/mailman/listinfo/python-announce-list Support the Python Software Foundation: http://www.python.org/psf/donations/
ANN: cmd2, an extenstion of cmd that parses its argument line
Dear all, I would like to announce the first public release of cmd2, an extension of the standard library's cmd with argument parsing, here: https://github.com/anntzer/cmd2. Cmd2 is an extension built around the excellent cmd module of the standard library. Cmd allows one to build simple custom shells using ``do_*`` methods, taking care in particular of the REPL loop and the interactive help. However, no facility is given for parsing the argument line (do_* methods are passed the rest of the line as a single string argument). With Cmd2, ``do_*`` methods are type-annotated, either using Python 3's function annotation syntax, or with an ad-hoc ``annotate`` decorator, allowing the dispatcher to parse the argument list for them. Antony Lee -- http://mail.python.org/mailman/listinfo/python-list
