Documentation of the wait flag values inconsistent with the code (?)

[GoogleCodeExporter]

I am having some problems understanding some of the values of the wait flag.

In particular, 
1) According to the documentation 
(https://code.google.com/p/ndt/wiki/NDTProtocol#Queuing_clients), the value of 
the wait flag can be N, where N is "the estimated number of minutes that the 
Client will wait for its test to begin".
2) AFAIU from the server code, the server sends the number of pending tests 
(https://code.google.com/p/ndt/source/browse/trunk/src/web100srv.c#114), where 
tests == test clients (i.e., each individual test  - s2c, c2s, ... - has its 
own client).
3) Comments of the Java and c clients say that, "number of minutes to wait = 
number of queued clients"
https://code.google.com/p/ndt/source/browse/trunk/Applet/Tcpbw100.java#2477
https://code.google.com/p/ndt/source/browse/trunk/src/web100clt.c#725

My interpretation is
- The server sends the number of waiting test clients.
- The client estimates the waiting time as 45sec * number of tests.
If my interpretation is correct, 
- All the code (c server, c client, Java client) is correct.
- While the comments and the wiki seem wrong.

Could you confirm if the comments and the wiki are inconsistent or if I am just 
misreading them?

Thanks!

Original issue reported on code.google.com by tizi...@google.com on 22 Nov 2013 at 4:34

[GoogleCodeExporter]

The server is sending the number of waiting test clients.

However, I think that the documentation is correct, but the c/Java clients are 
incorrectly treating this number.

The documentation is consistent with the c server code:

https://code.google.com/p/ndt/source/browse/trunk/src/web100srv.c#2421

Original comment by jslawin...@soldevelo.com on 25 Nov 2013 at 2:22

[GoogleCodeExporter]

I think treating N as "number of minutes" makes more sense than "number of 
tests in line ahead of you", though I'm fine with either interpretation so long 
as the client and server are consistent.

Original comment by AaronMat...@gmail.com on 26 Nov 2013 at 1:04

[GoogleCodeExporter]

If I understand correctly the c server code 
(https://code.google.com/p/ndt/source/browse/trunk/src/web100srv.c#2421), the 
expected behavior is to have the client wait Nx60sec. As a consequence, the 
wiki is correct, as N is both the number of clients and the number of minutes.

OTOH, both the Java client 
(https://code.google.com/p/ndt/source/browse/trunk/Applet/Tcpbw100.java#2483) 
and the c client 
(https://code.google.com/p/ndt/source/browse/trunk/src/web100clt.c#731)should 
be changed to wait for 'wait' * 60 (instead of 'wait' * 45).

Could you confirm that this is the case?

Thanks again!

Original comment by tizi...@google.com on 27 Nov 2013 at 8:01

[GoogleCodeExporter]

Yes, in my opinion we should treat N as "number of minutes", so we should 
update both clients to be consistent with the documentation and the server code.

Original comment by jslawin...@soldevelo.com on 28 Nov 2013 at 9:06

[GoogleCodeExporter]

Original comment by smale...@soldevelo.com on 13 Feb 2014 at 2:11

• Changed state: Started

[GoogleCodeExporter]

Changes available on branch applet_103. Feel free to review.

Original comment by smale...@soldevelo.com on 14 Feb 2014 at 10:09

• Changed state: Complete

[GoogleCodeExporter]

Seems fine to me

Original comment by AaronMat...@gmail.com on 14 Feb 2014 at 1:39

[GoogleCodeExporter]

Original comment by smale...@soldevelo.com on 14 Feb 2014 at 1:47

• Changed state: Fixed

[GoogleCodeExporter]

Original comment by skost...@soldevelo.com on 23 Jun 2014 at 5:52

• Added labels: Milestone-NDT-3.7.0