You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@tomcat.apache.org by Ken Cheung <ms...@gmail.com> on 2012/02/28 17:40:24 UTC
Inconsistency in AprEndpoint.java and JIoEndpoint.java
I observed some code clones in Tomcat and found inconsistent code. Could anyone explain why this is not a bug?
/tomcat/trunk/java/org/apache/tomcat/util/net/AprEndpoint.java
1048 try {
1049 Thread.sleep(1000);
1050 } catch (InterruptedException e) {
1051 // Ignore
1052 }
1053 long now = System.currentTimeMillis();
1054 Iterator<SocketWrapper<Long>> sockets =
1055 waitingRequests.iterator();
1056 while (sockets.hasNext()) {
1057 SocketWrapper<Long> socket = sockets.next();
1058 if (socket.async) {
1059 long access = socket.getLastAccess();
1060 if (socket.getTimeout() > 0 &&
1061 (now-access)>socket.getTimeout()) {
1062
processSocketAsync(socket,SocketStatus.TIMEOUT);
1063 }
1064 }
1065 }
/tomcat/trunk/java/org/apache/tomcat/util/net/JIoEndpoint.java
147 try {
148 Thread.sleep(1000);
149 } catch (InterruptedException e) {
150 // Ignore
151 }
152 long now = System.currentTimeMillis();
153 Iterator<SocketWrapper<Socket>> sockets =
154 waitingRequests.iterator();
155 while (sockets.hasNext()) {
156 SocketWrapper<Socket> socket = sockets.next();
157 long access = socket.getLastAccess();
158 if (socket.getTimeout() > 0 &&
159 (now-access)>socket.getTimeout()) {
160 processSocketAsync(socket,SocketStatus.TIMEOUT);
161 }
162 }
Quick description of the inconsistency
Two code snippets are very similar code, but as you see, in JIoEndpoint.java does not check "if (socket.async)" while AprEndpoint.java has the checker.
Re: Inconsistency in AprEndpoint.java and JIoEndpoint.java
Posted by Mark Thomas <ma...@apache.org>.
On 29/02/2012 04:51, Ken Cheung wrote:
> Because both of them are performing the same function. However, one
> of them (AprEndpoint.java:1057:1058) you typecast "SocketWrapper" to
> "Long" and check if "socket.async" is true while another one
> (JloEndpoint.java) you just typecast it to "Socket" and execute the
> same piece of codes. The checking of "socket.async" is missing here
> and therefore it could be a bug.
Oh dear. Where to start...?
Some things to consider:
- The code being different is no basis (on its own) for there being a
bug. The code snippets are supporting different Endpoint implementations
with different properties so it is entirely possible that the
implementations are different.
- Typecast? No. Not even close. You need to go and read up on generics
in Java.
- You assume that the check is missing in JIoEndpoint. Surely it is
equally possible that the check is unnecessary in AprEndpoint. You don't
appear to have considered that possibility.
- Before stating there is a bug, you need to understand what the code is
doing and have a credible explanation as to what might go wrong. You
don't have that. All you have is a statement that the code is different.
For the record, the code is correct and there is no bug but for the
benefit of the OP I'd like to see them work out why this is the case
rather than me spoon feeding them the explanation.
Mark
>
> On 29 Feb, 2012, at 2:45 AM, Mark Thomas wrote:
>
>> On 28/02/2012 16:40, Ken Cheung wrote:
>>> I observed some code clones in Tomcat and found inconsistent
>>> code. Could anyone explain why this is not a bug?
>>
>> Yes.
>>
>> Perhaps you'd like to explain your basis for assuming it is a bug.
>>
>> Mark
>>
>> ---------------------------------------------------------------------
>>
>>
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
>> For additional commands, e-mail: users-help@tomcat.apache.org
>>
>
>
> ---------------------------------------------------------------------
>
>
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
> For additional commands, e-mail: users-help@tomcat.apache.org
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
RE: Inconsistency in AprEndpoint.java and JIoEndpoint.java
Posted by "Caldarale, Charles R" <Ch...@unisys.com>.
> From: Ken Cheung [mailto:msrbugzilla@gmail.com]
> Subject: Re: Inconsistency in AprEndpoint.java and JIoEndpoint.java
> you typecast "SocketWrapper" to "Long" and check if "socket.async"
> is true while another one (JloEndpoint.java) you just typecast it
> to "Socket" and execute the same piece of codes.
If you think SocketWrapper<Long> is a type-cast, perhaps you need to do a bit more basic Java homework before telling people their code has bugs in it. (Hint: SocketWrapper<> is, as its name and syntax suggest, a generic that adds behavior while hiding - that is to say, wrapping - any specified type.)
> The checking of "socket.async" is missing here and therefore it could
> be a bug.
So could not checking if it's Tuesday, but that makes about as much sense. (Hint: read the docs; check to see if the Jio connector can use asynchronous sockets.)
- Chuck
THIS COMMUNICATION MAY CONTAIN CONFIDENTIAL AND/OR OTHERWISE PROPRIETARY MATERIAL and is thus for use only by the intended recipient. If you received this in error, please contact the sender and delete the e-mail and its attachments from all computers.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Inconsistency in AprEndpoint.java and JIoEndpoint.java
Posted by Ken Cheung <ms...@gmail.com>.
Because both of them are performing the same function. However, one of them (AprEndpoint.java:1057:1058) you typecast "SocketWrapper" to "Long" and check if "socket.async" is true while another one (JloEndpoint.java) you just typecast it to "Socket" and execute the same piece of codes. The checking of "socket.async" is missing here and therefore it could be a bug.
On 29 Feb, 2012, at 2:45 AM, Mark Thomas wrote:
> On 28/02/2012 16:40, Ken Cheung wrote:
>> I observed some code clones in Tomcat and found inconsistent code.
>> Could anyone explain why this is not a bug?
>
> Yes.
>
> Perhaps you'd like to explain your basis for assuming it is a bug.
>
> Mark
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
> For additional commands, e-mail: users-help@tomcat.apache.org
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Inconsistency in AprEndpoint.java and JIoEndpoint.java
Posted by Mark Thomas <ma...@apache.org>.
On 28/02/2012 16:40, Ken Cheung wrote:
> I observed some code clones in Tomcat and found inconsistent code.
> Could anyone explain why this is not a bug?
Yes.
Perhaps you'd like to explain your basis for assuming it is a bug.
Mark
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org