This issue tracker has been migrated to GitHub, and is currently read-only.
For more information, see the GitHub FAQs in the Python's Developer Guide.

classification
Title: asyncore does not react properly on close()
Type: behavior Stage: resolved
Components: Library (Lib) Versions: Python 2.7
process
Status: closed Resolution: not a bug
Dependencies: Superseder:
Assigned To: Nosy List: Gavin M. Roy, giampaolo.rodola, terry.reedy, tgeorgiev
Priority: normal Keywords:

Created on 2011-01-10 06:40 by tgeorgiev, last changed 2022-04-11 14:57 by admin. This issue is now closed.

Messages (10)
msg125882 - (view) Author: Teodor Georgiev (tgeorgiev) Date: 2011-01-10 06:40
I am trying to add a simple timer to each created socket and destroy it
once the timer expires:

class client(asyncore.dispatcher):
    def __init__(self,host):
    ...
    self.timeout = time.time() + 5

    def readable(self):
        if time.time() >= self.timeout:
            self.close()
        return True

When running that code, it raises an exception:

asyncore.loop(timeout=0.8)
  File "/usr/lib/python2.6/asyncore.py", line 211, in loop
    poll_fun(timeout)
  File "/usr/lib/python2.6/asyncore.py", line 144, in poll
    raise
  File "/usr/lib/python2.6/asyncore.py", line 141, in poll
    r, w, e = select.select(r, w, e, timeout)
select.error: (9, 'Bad file descriptor')


Although del_channel is executed properly and the socket is removed from the map, the poll function is not updated with that info and continues to keep the socket into the r,w,e.
msg125883 - (view) Author: Giampaolo Rodola' (giampaolo.rodola) * (Python committer) Date: 2011-01-10 07:35
Problem is you have to return False right after close().
There's an open issue to add a scheduler to asyncore:
http://bugs.python.org/issue1641
Closing this out as invalid.
msg125894 - (view) Author: Teodor Georgiev (tgeorgiev) Date: 2011-01-10 11:03
Sorry, I forgot to mention - I have already tried to return False, but there was no difference.

def readable(self):
        if time.time() >= self.timeout:
            self.close()
            return False
         else:
            return True
msg125895 - (view) Author: Giampaolo Rodola' (giampaolo.rodola) * (Python committer) Date: 2011-01-10 11:21
What if you return False also in writable method?
msg125897 - (view) Author: Teodor Georgiev (tgeorgiev) Date: 2011-01-10 11:44
Precisely, I traced down the problem by putting a simple "breakpoint"
in asyncore.py:

def poll(timeout=0.0, map=None):
    if map is None:
        map = socket_map
    if map:
        r = []; w = []; e = []
        for fd, obj in map.items():
            is_r = obj.readable()
            print "Readable??? -->" , is_r
            is_w = obj.writable()
            if is_r:
                r.append(fd)
            if is_w:
                w.append(fd)
            if is_r or is_w:
                e.append(fd)
        if [] == r == w == e:
            time.sleep(timeout)
            return

        print r,w,e
try:
            r, w, e = select.select(r, w, e, timeout)
        except select.error, err:
            if err.args[0] != EINTR:
                raise
            else:
                return



And here it comes:

[5] [5] [5]
Readable??? --> True
[5] [5] [5]
Readable??? --> True
[5] [5] [5]
Readable??? --> False
[] [5] [5]

Traceback (most recent call last):
  File "./dlms_client.py", line 136, in <module>
    asyncore.loop(timeout=0.8)
  File "/usr/lib/python2.6/asyncore.py", line 213, in loop
    poll_fun(timeout)
  File "/usr/lib/python2.6/asyncore.py", line 146, in poll
    raise
  File "/usr/lib/python2.6/asyncore.py", line 143, in poll
    r, w, e = select.select(r, w, e, timeout)
select.error: (9, 'Bad file descriptor')

So, in order this to work, on first sight all r,w,e must not point to
a socket that has been already closed. Now I am going to think for a workaround at least.
msg127149 - (view) Author: gmr (Gavin M. Roy) Date: 2011-01-26 21:55
What I noticed in tracing through the code is that it's getting stuck in a loop because it depends on grabbing asyncore.socket_map if map is null when passed into asyncore.loop. I got around this by appending:

asyncore.loop(0.1, map=[], count=1)

After my close(). I believe this bypasses the grab of socket_map from asyncore.socket_map and allows the while loop in asyncore.loop to exit cleanly.
msg127150 - (view) Author: gmr (Gavin M. Roy) Date: 2011-01-26 22:09
For more clarity, I am passing in a list because it will evaluate as False in the while loop on 209 and 213. The fix is to add len(map) to the while loops on those lines.
msg183760 - (view) Author: Terry J. Reedy (terry.reedy) * (Python committer) Date: 2013-03-08 19:47
Teodor or Gavin: is the (mis)behavior the same in 3.3?

Giampaolo: has the OP identified a fixable misbehavior relative to the documented behavior, making this a valid behavior issue?
Or is this instead an enhancement request, possibly superseded by #1641?
msg183761 - (view) Author: Giampaolo Rodola' (giampaolo.rodola) * (Python committer) Date: 2013-03-08 19:59
I'm not sure what the OP and Gavin are complaining about in their last messages. Could you guys be more clear and/or provide a code sample which reproduces the problem?
msg194007 - (view) Author: Terry J. Reedy (terry.reedy) * (Python committer) Date: 2013-07-31 18:54
I am re-closing as there is no defined bug to fix, and 2.7 is closed to enhancements.
History
Date User Action Args
2022-04-11 14:57:11adminsetgithub: 55087
2013-07-31 18:54:34terry.reedysetstatus: open -> closed
type: behavior
messages: + msg194007

resolution: not a bug
stage: resolved
2013-03-08 19:59:07giampaolo.rodolasetmessages: + msg183761
2013-03-08 19:47:57terry.reedysetnosy: + terry.reedy
messages: + msg183760
2011-06-01 06:24:59terry.reedysetversions: - Python 2.6, Python 2.5
2011-01-26 22:09:44Gavin M. Roysetnosy: giampaolo.rodola, tgeorgiev, Gavin M. Roy
messages: + msg127150
2011-01-26 21:55:15Gavin M. Roysetnosy: + Gavin M. Roy

messages: + msg127149
versions: + Python 2.5
2011-01-10 11:44:31tgeorgievsetnosy: giampaolo.rodola, tgeorgiev
messages: + msg125897
2011-01-10 11:21:05giampaolo.rodolasetnosy: giampaolo.rodola, tgeorgiev
messages: + msg125895
2011-01-10 11:03:20tgeorgievsetstatus: closed -> open

messages: + msg125894
resolution: not a bug -> (no value)
nosy: giampaolo.rodola, tgeorgiev
2011-01-10 07:35:24giampaolo.rodolasetstatus: open -> closed

nosy: + giampaolo.rodola
messages: + msg125883

resolution: not a bug
2011-01-10 06:40:22tgeorgievcreate