Error with notification count

[#1]

hey @jab1 sorry for the late response.

The last I heard was on Xmas Eve was that the issue was impacting multiple communities and was still being investigated. 

I'll drop them a chaser email now, but it wouldn't surprise me if they were still in change freeze for the holidays. 

[#2]

Fair enough, @JonoH . But I am amazed that such a bad bit of coding was allowed out. Had it only affected Plusnet, because of the 'tweaks' on here, I could say 'it happens', but to affect others as well points to poor testing - IMHO.

[#3]

They've responded to me already this morning 

"Unfortunately, we do not have any updates from our engineering team as of now in regards to the bug report filed by us. We have already requested them to provide us with some updates in regards to this. We will get back to you as soon as we hear back from them. Your patience is much appreciated."

[#4]

---

@jab1 wrote:

... I am amazed that such a bad bit of coding was allowed out. Had it only affected Plusnet, because of the 'tweaks' on here, I could say 'it happens', but to affect others as well points to poor testing - IMHO.

---

Hi John,

Whilst understanding your frustration, it is very challenging to test every bit of complex software, especially cases such as this.  from the profiling done by the community, it rather looks as the following ducks need to be neatly lined up to experience this issue...

1. The user has to be logged in with "keep the session alive"
2. The user has to have left a browser window open on the forums on a page for a period - believed to be longer than the website's keep alive timeout period
3. The user has to then refresh the page or navigate to somewhere other than notifications
4. Depending on some unknown factor users may see a non-specific notification error count variously reported as 101 and 47

Ducks 1-3 being aligned does not always cause #4 to appear.  The issue appears to be client side, not server side.  In my last post here I said that I would test using the SILK browser on my new (Christmas gift) kindle tablet.  I saw the issue there too, so it looks to be pan browser, however within that testing activity, I took two clients in states 1-2 through 3/4 ... both reported a 101 notification count, even though the first client went all the way through to looking at the notifications page BEFORE the second did duck #3.

The reality is that users will invariably find ways of breaking software by using in a manner never conceived by the developers - testing every possible logic route in every conceivable circumstance is not going to be achieved in a systems test lab ... I've run a few test programmes and I have seen users (ab)use my own code too.  One is left with the feeling "That was never part of the design ... but I see how the user got there ... and I have to accept it does not work ... but dammed if I can see why!".  If I were a gambler, I would be putting money on some third party component here.

[#5]

I didn't get the '101' this morning, just a single '1' notification which was correct when I checked.

[#6]

Some more intel in the hope it might help.

Had left my session exiting from a post I wanted to abandon, but had not actually clicked the LEAVE DISCARD CHANGES browser warning.

Clicked LEAVE and got a Oops! authentication error.

Top right status reported ONE notification.

Bottom left notification button count was 101.

Actual notifications were 8.

Event timed around 10:00 GMT today.

[#7]

@JonoH Is it my imagination or was this issue 'fixed' overnight?

[#8]

@Mook Not for me, it wasn't.

[#9]

Still broken for me. I had already cleared my notifications this morning, got the "1" due to a post here, then it jumped back up to 36 again!.

[#10]

Ah! Yes I think you're both right, it looks like it was nothing more than circumstance. 

[#11]

Hi @Mook  it's not been fixed I'm afraid, our vendors still tell us their engineers have identified the bug, but they've given us no further information.

[#12]

Thanks for the replay @JonoH but after the replies I got and then later in the day this was confirmed to me. But at least they are aware of what the issue, pity there's no time line to a fix, oh well.

[#13]

I've had an Email from the vendor today advising that they're still looking into the cause of this issue as it's impacting multiple Communities. I've escalated it further and will update you when I have more into

[#14]

Is this still being investigated?

[#15]

Don't underestimate the users, we're very adaptable : 101=0