strange behaviours when writing large amount of data on stdout

[manlio]

Regards.

In the file submitted there are all the instructions to reproduce 2 bugs and 1
strange behaviour.
I hope they are clear.

Python 2.4.1 (#65, Mar 30 2005, 09:13:57) [MSC v.1310 32 bit (Intel)] on win32
Twisted 2.0.1

Thanks for all your work!

[spiv]

Interesting; this doesn't reproduce on linux, so it appears to be Windows 
specific.

I'm not sure what this part of your script is about, by the way:
    d = {
        1: [['DATA', ['XXX', 'YYY'], [data]]],
        2: [['DATA', ['ZZZ', 'WWW'], [data]]]
        }
The 'd' variable isn't ever referenced.

Anyway, I'm assigning to justinj, because he's the lucky guy that's been fixing
our win32 issues lately :)

justinj, if you don't want to deal with this, assign back to me and I'll try
hassling someone else.

Thanks for the report, and especially for the script to reproduce!

[manlio]

At 12.29 02/11/2005 +0000, you wrote:

>Andrew Bennetts <andrew-twistedbugs@puzzling.org> added the comment:
>
>Interesting; this doesn't reproduce on linux, so it appears to be Windows 
>specific.
>
>I'm not sure what this part of your script is about, by the way:
>    d = {
>        1: [['DATA', ['XXX', 'YYY'], [data]]],
>        2: [['DATA', ['ZZZ', 'WWW'], [data]]]
>        }
>The 'd' variable isn't ever referenced.

It is my fault!
I found this bug playing with IMAP4Client.fetch, so I was trying to reproduce that data structure.
Just forget about it.

>Anyway, I'm assigning to justinj, because he's the lucky guy that's been fixing
>our win32 issues lately :)

Ok, thanks.
I hope this bug is not specific to my PC.
Anyway, I'm using Windows XP

>justinj, if you don't want to deal with this, assign back to me and I'll try
>hassling someone else.
>
>Thanks for the report, and especially for the script to reproduce!

Thanks to you!

Regards  Manlio Perillo

[justinj]

Manlio, for cases 1 and 2 listed in your attached file do you still see the 
problem when running with the latest code in svn trunk?  Case 3 seems like we 
just need to handle interrupts more cleanly, but I haven't looked into it.

[manlio]

I have now installed Twisted from trunk, but I still have the problems.

Did someone try to reproduce the issue on a Windows system?

[justinj]

Yes, I ran the script following the guidelines in the comments and couldn't 
reproduce problem 1 and 2.

[manlio]

Ok.

I have started a new thread on comp.lang.python:
http://groups.google.com/group/comp.lang.python/browse_frm/thread/84e6749132529a40/daa24777bfaf3463#daa24777bfaf3463

with the minimal code causing the problem:
IOError: [Errno 12] Not enough space


Let's wait for someone who can reproduce the bug...

[justinj]

Closing since it is clear the bug is not related to twisted and no one else 
can reproduce it.

[manlio]

Well, it seems that this is a problem caused by a bug on Windows XP without
Service Pack.

Now I have installed SP2 and the bug is gone...

[Jean-Paul Calderone]

It seems as though this is still a problem on WinXP without SP2.  Perhaps we do
not have the manpower to devote to fixing this right now, but that does not make
the bug invalid.  It is good to have things like this reported, and good to use
the tracker to, er, track them; perhaps someone will even notice the issue and
fix it for us.

[jknight]

The exception (issue #3) is certainly worrying and quite possibly a bug, but I 
don't think we really need to be tracking the misbehavior of "print" in python 
only on Windows XP < SP2. If you're that worried about it, probably forwarding it 
to the python bug tracker would be a better idea.

[manlio]

As I can think the exception 
exceptions.IOError: [Errno 12] Not enough space
is a Windows bug, Python can do nothing about it.

The problem is that it causes Python to crash, when (and only when?) using Twisted.

[Marcin Kasperski]

I am getting the 'unsubscriptable object' error on Linux (debian, python-twisted-core 2.4.0-2). The error happens while debugging, when I Stop debugging. A bit more info about the error context (maybe something of this will give somebody some idea...):

• I have twisted app (low level) which uses default reactor and connects to it one client telnet (reactor.connectTCP with some ReconnectingClientFActory subclass object as param) and one server telnet (reactor.listenTCP with simple ServerFactory subclass object as param) and also uses adbapi

• the client telnet has some connection problems (connects but is soon dropped) so it repeats his attempts every few seconds

• at some moment I click STOP in debugger, then I got the error with the following backtrace:

(....)
            reactor.run()
          File "/usr/lib/python2.4/site-packages/twisted/internet/posixbase.py",                 line 218, in run
            self.mainLoop()
          File "/usr/lib/python2.4/site-packages/twisted/internet/posixbase.py",                 line 226, in mainLoop
            self.runUntilCurrent()
        --- <exception caught here> ---
          File "/usr/lib/python2.4/site-packages/twisted/internet/base.py", line                 555, in runUntilCurrent
            call.func(*call.args, **call.kw)
          File "/usr/lib/python2.4/site-packages/twisted/internet/base.py", line                 411, in _continueSystemEvent
            for callList in sysEvtTriggers[1], sysEvtTriggers[2]:
        exceptions.TypeError: unsubscriptable object

The context (reconnectingclientfactory repeatable trying) can be important, I have never seen this error before I started getting those connection problems. It can be that state of something is somehow broken at such stage...

[Jean-Paul Calderone]

I don't think the previous comment has anything to do with this ticket. It seems to be an issue with calling reactor.stop more than once.

[Rotund]

Updated INSTALL to document SP2 or higher is required.

[Rotund]

Update to INSTALL

[TimAllen]

I like this documentation patch, but when you say which platforms are supported, you should mention the version numbers for Debian, Ubuntu and Fedora, as you do for OS X and Windows. In fact, it's probably more important, since Linux tends to change more rapidly than other OSs.

[Jean-Paul Calderone]

For the purposes of record keeping, if we do choose to go this route, let's close this ticket as wontfix and open a new one (or find an existing one?) about documenting Twisted's platform requirements. Thanks.

[Jonathan Lange]

This didn't make the 10.0 release. I also don't think it should block the 10.1 release.

[Jean-Paul Calderone]

Similar-looking Python bug report, ​http://bugs.python.org/issue11395

[Rotund]

Updated with Linux release versions and for new file

[Rotund]

Updated the patch. Up to date with trunk and added the linux versions.

[Rotund]

Also updated ReleaseProcess in wiki and pointed the update to look for currently supported versions and have instruction pending this ticket.

[Moshe Zadka]

Looks good

[Jean-Paul Calderone]

From the progress on the Python issue, it looks like this may not be limited to XP < SP2.

Also, in my earlier comment, I asked for a separate ticket for the doc change. Maybe that was wrong, but don't just ignore it, explain why you disagree.

[Jean-Paul Calderone]

I have almost no motivation to do any work on this ticket.

[hawkowl]

This is a Windows problem which has been fixed in upstream Python.