Github user canselcik commented on the pull request:
https://github.com/apache/trafficserver/pull/323#issuecomment-153974633
How about `ClassAllocator spdyRequestAllocator`, though? If
the right fix is to remove `~SpdyClientSession()`, then should we also remove
`~SpdyRequest()`?
Github user bgaff commented on the pull request:
https://github.com/apache/trafficserver/pull/323#issuecomment-153956757
Cool, sounds like we're on the same page @jpeach.
@canselcik, you also cool w/ this proposal?
---
If your project is set up for it, you can reply to this
Github user jpeach commented on the pull request:
https://github.com/apache/trafficserver/pull/323#issuecomment-153956478
Ah I see now. Then probably the right fix is to remove
```~SpdyClientSession()```. I don't think any other proxy-allocated object have
destructors, since that can'
Github user bgaff commented on the pull request:
https://github.com/apache/trafficserver/pull/323#issuecomment-153956235
@jpeach , given that we follow an init() / clear() pattern anyway, what are
your thoughts on that actually: just remove the call to clear() from the
destructor.
-
Github user bgaff commented on the pull request:
https://github.com/apache/trafficserver/pull/323#issuecomment-153955267
these objects should _never_ go out of scope until the process dies (they
live on a freelist), so isn't it enough just to do nothing in the destructor?
---
If your
Github user canselcik commented on the pull request:
https://github.com/apache/trafficserver/pull/323#issuecomment-153954924
Upon exit `static ClassAllocator
spdyClientSessionAllocator` goes out of scope and its destructor is called,
which goes on to call the destructor on the `SpdyCl
Github user jpeach commented on the pull request:
https://github.com/apache/trafficserver/pull/323#issuecomment-153951633
Yeh that's the way these object usually work. ```SpdyClientSession``` is a
bit different from the other client session objects because last time I cleaned
this up
Github user canselcik commented on the pull request:
https://github.com/apache/trafficserver/pull/323#issuecomment-153950655
@jpeach `ClassAllocator`, when initialized, creates an
instance of `SpdyClientSession` and puts it aside. At this point, for this
"prototype", `SpdyClientSessio
Github user jpeach commented on the pull request:
https://github.com/apache/trafficserver/pull/323#issuecomment-153950616
But the only way this could happen is if the session was released before
```init``` was called, right?
---
If your project is set up for it, you can reply to this
Github user bgaff commented on the pull request:
https://github.com/apache/trafficserver/pull/323#issuecomment-153950509
@jpeach the only way to deal with that would be a mutex that protects
initialization and destruction. The problem is that these objects are
constructed via static i
Github user jpeach commented on the pull request:
https://github.com/apache/trafficserver/pull/323#issuecomment-153949955
How do we end up freeing a partially-initialized ```SpdyClientSession```?
It looks like this does not fix the root cause of the problem.
---
If your project is se
Github user bgaff commented on the pull request:
https://github.com/apache/trafficserver/pull/323#issuecomment-153948633
I don't see an issue w/ this, I'll merge it if no one else has issues.
---
If your project is set up for it, you can reply to this email and have your
reply appear
GitHub user canselcik opened a pull request:
https://github.com/apache/trafficserver/pull/323
Dereferencing a NULL pointer in SpdyClientSession::clear()
- At `SpdyClientSession.cc:28`, `static ClassAllocator
spdyClientSessionAllocator` creates an instance of `SpdyClientSession` us
Github user asfgit closed the pull request at:
https://github.com/apache/trafficserver/pull/322
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature
GitHub user SolidWallOfCode opened a pull request:
https://github.com/apache/trafficserver/pull/322
TS-306: Fix file privilege elevation to work with log rotation.
This fixes several issues.
The biggest is DAC bypass privilege elevation. Because `traffic_server` is
invoked
Github user danobi commented on the pull request:
https://github.com/apache/trafficserver/pull/320#issuecomment-153824370
Looks good to me
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this
Github user jpeach commented on the pull request:
https://github.com/apache/trafficserver/pull/321#issuecomment-153798201
I think that we really need to break this feature down into smaller commits
before we can digest and merge it. Right now the only way I can see doing it is
if we s
Github user SolidWallOfCode commented on the pull request:
https://github.com/apache/trafficserver/pull/320#issuecomment-153797825
Dan, here's my current version of this PR -
https://github.com/SolidWallOfCode/trafficserver/tree/pr/320
I have to head to lunch now.
---
If your pro
Github user SolidWallOfCode commented on a diff in the pull request:
https://github.com/apache/trafficserver/pull/320#discussion_r43909560
--- Diff: proxy/shared/DiagsConfig.cc ---
@@ -281,7 +281,8 @@ DiagsConfig::DiagsConfig(const char *filename, const
char *tags, const char *act
Github user SolidWallOfCode commented on a diff in the pull request:
https://github.com/apache/trafficserver/pull/320#discussion_r43908535
--- Diff: proxy/logging/Log.cc ---
@@ -1242,7 +1242,15 @@ Log::flush_thread_main(void * /* args ATS_UNUSED */)
break;
Github user SolidWallOfCode commented on a diff in the pull request:
https://github.com/apache/trafficserver/pull/320#discussion_r43908399
--- Diff: proxy/shared/DiagsConfig.cc ---
@@ -281,7 +281,8 @@ DiagsConfig::DiagsConfig(const char *filename, const
char *tags, const char *act
GitHub user jrushf1239k opened a pull request:
https://github.com/apache/trafficserver/pull/321
Parent selection 2.0
This is a refactor of Parent Selection to add the following features:
1) multi-site origin - using parent selection, origin servers may be
selected using rou
Github user SolidWallOfCode commented on the pull request:
https://github.com/apache/trafficserver/pull/320#issuecomment-153761311
Reviewing now. Looks good overall.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your
23 matches
Mail list logo
