[Log4cplus-devel] [log4cplus:bugs] #318 socket copy operator closes the socket

[Václav H. <wi...@us...>]

The problem here is that historically, `Socket s(rhs)` would move data from `rhs` to `s`. This is not good but I am not inclined to change this. This is fixed on master branch by proper implementation of move constructor and move assignment operator.

Thereofre I am closing this as wront-fix for 1.2.x branch.


---

** [bugs:#318] socket copy operator closes the socket**

**Status:** open
**Group:** v1.2.x
**Labels:** socket 
**Created:** Sun May 31, 2015 08:14 AM UTC by Alexis Wilke
**Last Updated:** Sun May 31, 2015 09:02 AM UTC
**Owner:** Václav Haisman
**Attachments:**

- [0002-log4cplus-1.2.0-rc3-socket-copy-operator.patch](https://sourceforge.net/p/log4cplus/bugs/318/attachment/0002-log4cplus-1.2.0-rc3-socket-copy-operator.patch) (262 Bytes; text/x-patch)


As I'm working on the simpleserver to make it work properly (especially, so it does not leak all the ClientThread objects--this will be another issue), I noticed that the copy constructor properly transferred a socket, whereas, the copy assignment closes the socket!

So if I do this it works as expected:

    Socket s(rhs);  // rhs transferred to s (now rhs is considered closed)

Whereas, doing this closes the socket:

    Socket s;
    s = rhs;  // s is a closed socket (So is rhs)

I think that in both cases we should expect the same behavior. There is the function that does it wrong. I suggest to remove the call to the close() function:

    AbstractSocket&
    AbstractSocket::operator=(const AbstractSocket& rhs)
    {
        if(&rhs != this) {
            close();    // <-- this should not happen!
            copy(rhs);
        }

        return *this;
    }

I'm attaching a patch.




---

Sent from sourceforge.net because log...@li... is subscribed to https://sourceforge.net/p/log4cplus/bugs/

To unsubscribe from further messages, a project admin can change settings at https://sourceforge.net/p/log4cplus/admin/bugs/options.  Or, if this is a mailing list, you can unsubscribe from the mailing list.