Re: SocketAddress.toString() and RFC 2732

2019-06-07 Thread Alan Bateman
On 07/06/2019 19:46, Chris wrote: java.net.SocketAddress can represent an IPv6 address with port. Because parts of an IPv6 address are separated with colons and a colon is also used to separate the address from the port, RFC 2732 specifies that the address be in square brackets in this case, e.g.

Re: RFR [13] 8225426: Replace plain with system-default in Socket and ServerSocket

2019-06-07 Thread Alan Bateman
On 07/06/2019 15:16, Chris Hegarty wrote: : “socket implementation factory” is better. Updated: Looks good, many inconsistencies resolved. For the CSR then the only real change is the no-arg Socket constructor to specify existing behavior, everything else is below the radar. -Alan

Re: RFR [13] 8225426: Replace plain with system-default in Socket and ServerSocket

2019-06-07 Thread Chris Hegarty
> On 7 Jun 2019, at 15:02, Alan Bateman wrote: > > > > On 07/06/2019 14:20, Chris Hegarty wrote: >> Alan, Daniel, >> >>> On 7 Jun 2019, at 13:07, Alan Bateman wrote: >>> >>> On 07/06/2019 12:24, Chris Hegarty wrote: ... >>> Replacing with "plain" with system-default is okay but it do

Re: RFR [13] 8225426: Replace plain with system-default in Socket and ServerSocket

2019-06-07 Thread Alan Bateman
On 07/06/2019 14:20, Chris Hegarty wrote: Alan, Daniel, On 7 Jun 2019, at 13:07, Alan Bateman wrote: On 07/06/2019 12:24, Chris Hegarty wrote: ... Replacing with "plain" with system-default is okay but it does remind me of problems in the other constructors. The no-arg Socket constructo

Re: RFR [13] 8225426: Replace plain with system-default in Socket and ServerSocket

2019-06-07 Thread Daniel Fuchs
On 07/06/2019 14:20, Chris Hegarty wrote: Complete set of changes ( will be copied as-is into the CSR ): --- LGTM Chris! best regards, -- daniel

Re: RFR [13] 8225426: Replace plain with system-default in Socket and ServerSocket

2019-06-07 Thread Chris Hegarty
Alan, Daniel, > On 7 Jun 2019, at 13:07, Alan Bateman wrote: > > On 07/06/2019 12:24, Chris Hegarty wrote: >> ... > Replacing with "plain" with system-default is okay but it does remind me of > problems in the other constructors. The no-arg Socket constructor is > currently specified to use a

Re: RFR [13] 8225426: Replace plain with system-default in Socket and ServerSocket

2019-06-07 Thread Alan Bateman
On 07/06/2019 12:24, Chris Hegarty wrote: Another minor followup that we agreed to address after the integration of 8221481 (Reimplement the Legacy Socket API). Prior to 8221481, there was just a single default SocketImpl, the "plain" socket impl. There are now two, PlainSocketImpl and NioSocket

Re: RFR [13] 8225426: Replace plain with system-default in Socket and ServerSocket

2019-06-07 Thread Daniel Fuchs
Hi Chris, looks good to me. I just wonder if: "Otherwise a system-default implementation is created." would read better - but I'll defer to your judgement. I also noticed one typo: On 07/06/2019 12:24, Chris Hegarty wrote: [snip] src/java.base/share/classes/java/net/ServerSocket.java [s

[teststabilization] RFR: 8224865: Replace wildcard address with loopback or local host in tests - part 13

2019-06-07 Thread Daniel Fuchs
Hi, Please find below a fix for: 8224865: Replace wildcard address with loopback or local host in tests - part 13 https://bugs.openjdk.java.net/browse/JDK-8224865 webrev: http://cr.openjdk.java.net/~dfuchs/webrev_8224865/webrev.00/ Additional changes to note: Some tests (B6401598.jav

RFR [13] 8225426: Replace plain with system-default in Socket and ServerSocket

2019-06-07 Thread Chris Hegarty
Another minor followup that we agreed to address after the integration of 8221481 (Reimplement the Legacy Socket API). Prior to 8221481, there was just a single default SocketImpl, the "plain" socket impl. There are now two, PlainSocketImpl and NioSocketImpl, the latter being the default. The fo