Excellent!  I had made some progress starting to track this down but hadn’t 
gotten as far as you.
I’ll pull and git it a go.
— Dale

> On Sep 20, 2017, at 7:55 AM, Christofer Dutz <christofer.d...@c-ware.de> 
> wrote:
> 
> Hi all,
> 
> ok so finding this little detail got me on the right path. I updated index.js 
> to use “source” instead of “sourceIdx” and “target” instead of “targetIdx” 
> and now the console is working identically to the original gradle version.
> 
> Chris
> 
> 
> 
> Am 20.09.17, 13:30 schrieb "Christofer Dutz" <christofer.d...@c-ware.de>:
> 
>    Ok so today I investigated the problem a little more.
> 
>    Seems to be one of those cool problems you get when operating with a type 
> less language ;-)
> 
>    There is a difference between the original Sankey.js and the Edgent 
> version. The difference is that the original uses “source” and “target” while 
> the Edgent version uses “sourceIdx” and “targetIdx” so I am getting 
> JavaScript errors in the browser. 
>    I think I should try and find a solution for this and then see if the 
> charts come back. If that doesn’t help I can still get my hands dirty by 
> digging in the depths of the system.
> 
>    Chris
> 
> 
> 
>    Am 19.09.17, 17:59 schrieb "Susan Cline" <home4...@pacbell.net>:
> 
>        I did look at Chris’s pull request and don’t see any changes to the 
> file that is formatting the metrics.
> 
>        I also looked at the java class that is sending the metrics to the 
> console - that code looks unchanged as well, so I am at a loss for what did 
> change.
>        Perhaps some of the underlying java classes the servlets or utility 
> classes accessed changed?
> 
>        Susan
> 
> 
>> On Sep 18, 2017, at 8:02 AM, Dale LaBossiere <dml.apa...@gmail.com> wrote:
>> 
>> Hmm… I’m not familiar with that code but will take a look.
>> 
>> Susan, any thoughts, including whether or not this sort of difference could 
>> account for the graph not getting rendered but the console page otherwise 
>> looking ok?  (reminder this is pr309 and you can clone Chris’s fork to get 
>> all the changes)
>> 
>> — Dale
>> 
>>> On Sep 18, 2017, at 8:54 AM, Christofer Dutz <christofer.d...@c-ware.de> 
>>> wrote:
>>> 
>>> Hi all,
>>> 
>>> well I did have a more detailed look at the original (Gradle) version and 
>>> the maven version. When running both examples, I could see two requests 
>>> every 5 seconds. Both of them are answered by the server. The “jobs” 
>>> request is answered identically. The difference seems to be in the 
>>> “metrics” response:
>>> 
>>> The working version of the Gradle version produces this:
>>> 
>>> {
>>>  "jobId": "JOB_0",
>>>  "ops": [
>>>      {
>>>          "opId": "OP_2",
>>>          "metrics": [
>>>              {
>>>                  "type": "counter",
>>>                  "name": "Count",
>>>                  "value": "219"
>>>              }
>>>          ]
>>>      }
>>>  ]
>>> }
>>> 
>>> The Maven version however produces this:
>>> 
>>> {
>>>  "jobId": "JOB_0",
>>>  "ops": [
>>>      {
>>>          "opId": "OP_2",
>>>          "metrics": [
>>>              {
>>>                  "type": "long",
>>>                  "name": "Count",
>>>                  "value": "436"
>>>              }
>>>          ]
>>>      }
>>>  ]
>>> }
>>> 
>>> The difference seems to be that the “type” is “counter” in the working 
>>> version and “long” in the not working version. I tried debugging this, but 
>>> I couldn’t really understand why this is different.
>>> 
>>> Any ideas what could be wrong? I guess in general things should be working, 
>>> all we have to do is fix some final mini-quirks.
>>> 
>>> Chris
>> 
> 
> 
>        .
> 
> 
> 
> 

Reply via email to