Re: [Twisted-Python] HTTP Agent persistent connections not closed for some HTTPS sites

2017-03-07 Thread Cory Benfield

> On 6 Mar 2017, at 09:05, Adi Roiban  wrote:
> 
> I am not sure about which code are we talking here.
> The specific HTTP11ClientProtocol which will fix only the HTTP client
> part or the generic TLSMemoryBIOProtocol code which might fix any TLS
> connection?

I’m talking about the fact that calling TLSMemoryBIOProtocol.loseConnection may 
not ever actually drop the underlying FD is the surprising part of the code. We 
had to work around it in the server side by adding a timeout after 
loseConnection is called to end up calling abortConnection.

Cory___
Twisted-Python mailing list
Twisted-Python@twistedmatrix.com
http://twistedmatrix.com/cgi-bin/mailman/listinfo/twisted-python


Re: [Twisted-Python] non-merge commits to trunk & "review" keyword

2017-03-07 Thread Tristan Seligmann
On Tue, 7 Mar 2017 at 09:56 Glyph Lefkowitz  wrote:

> On Mar 6, 2017, at 9:02 PM, Tristan Seligmann 
> wrote:
>
> On Tue, 7 Mar 2017 at 06:38 Glyph Lefkowitz 
> wrote:
>
>
> This is definitely bad, forbidden by existing policy, etc.  In fact I
> remember adjusting the settings so that the 'merge' button would always
> create a merge commit; in fact, the configuration is still set that way.
>
>
> Rebase merging was allowed when I checked earlier, so I disabled it (I
> guess I managed to do this before you looked at the configuration).
>
>
> Perhaps I only did this for some other repos, then.  Did I miss your
> announcement of this?  Hint, hint? :)
>

 By the time I got to my email client (which admittedly was several hours
after I made the configuration change), your email had arrived, so my reply
ended up being the "announcement" too. The last time I looked at these
settings, rebasing was not allowed, so I guess somebody else turned it on
(possibly by accident); the GitHub audit logs don't go far back enough for
me to check who did it and when, but that means the change happened over
several months ago.

> I suspect that is what happened here, but note also that these settings
> only control what is possible by pressing the merge button on GitHub; you
> can construct commits locally however you like and push them, as long as
> you get green commit statuses for all the mandatory commit checks.
>
>
> Is this true even for people with just repo:write?
>

I haven't actually tested this scenario, so I could be mistaken; inferring
from other things it shouldn't matter what your permissions are (for a
non-force push).
___
Twisted-Python mailing list
Twisted-Python@twistedmatrix.com
http://twistedmatrix.com/cgi-bin/mailman/listinfo/twisted-python


Re: [Twisted-Python] non-merge commits to trunk & "review" keyword

2017-03-07 Thread Craig Rodrigues
On Mon, Mar 6, 2017 at 8:37 PM, Glyph Lefkowitz 
wrote:

>
>
> Craig, since you're the one who made this merge, can you explain what
> happened?  Has github's 'merge' button stopped prompting for a commit
> message?  Failing to wait for the removal of the 'review' keyword was
> obviously a mistake, but based on my previous experience with the 'merge'
> button this wouldn't even be a type of mistake that's _possible_ to make;
> understanding what you did to trigger it would therefore be very useful.
>
>
Woops, sorry about that.  I pressed the Merge button in git, but I guess I
clicked on the wrong
option after that.  It is good that this is now disabled to prevent future
mistakes like this.

--
Craig
___
Twisted-Python mailing list
Twisted-Python@twistedmatrix.com
http://twistedmatrix.com/cgi-bin/mailman/listinfo/twisted-python


Re: [Twisted-Python] non-merge commits to trunk & "review" keyword

2017-03-07 Thread Glyph Lefkowitz

> On Mar 7, 2017, at 10:40 AM, Craig Rodrigues  wrote:
> 
> 
> 
> On Mon, Mar 6, 2017 at 8:37 PM, Glyph Lefkowitz  > wrote:
> 
> 
> Craig, since you're the one who made this merge, can you explain what 
> happened?  Has github's 'merge' button stopped prompting for a commit 
> message?  Failing to wait for the removal of the 'review' keyword was 
> obviously a mistake, but based on my previous experience with the 'merge' 
> button this wouldn't even be a type of mistake that's _possible_ to make; 
> understanding what you did to trigger it would therefore be very useful.
> 
> 
> Woops, sorry about that.  I pressed the Merge button in git, but I guess I 
> clicked on the wrong
> option after that.  It is good that this is now disabled to prevent future 
> mistakes like this.

OK, good to know; so this was a repo misconfiguration coupled with human error. 
 Glad we have something in place to prevent it now :).

Craig - that said, can you go back and fix up the relevant state (kill the 
'review' keyword, close the ticket with a reference to the relevant commit, and 
so on)?

-glyph

___
Twisted-Python mailing list
Twisted-Python@twistedmatrix.com
http://twistedmatrix.com/cgi-bin/mailman/listinfo/twisted-python