1 Topic by Syrphus

  • Syrphus
  • Veteran
  • Offline
  • From: Strathpeffer, Highland
  • Registered: 16-12-2008
  • Posts: 535

Topic: Yes, you've guessed it.

0811h, G4 working OK, G5 not showing grid maps and the IMT is covered in pink squares (tested in Opera and Chrome).

We are now over 6 months from the much-fanfared launch of G5 that was going to be the greatest thing in biological data presentation ever, would eliminate all the performance problems that were evident in G4, and (for all I know) bring everlasting peace and love to the world.

Instead, we have a dreadfully designed, fundamentally unstable mess that I don't believe has functioned properly for any consecutive 5 days since the launch.  It is obvious to me and many others that the problem goes right to the fundamental design and structure, and the reason we are in this mess reflects a complete lack of effective testing of the systems during development.

When there is a problem, it seems that someone tweaks something in the hope that it will improve the situation, but without knowing what the real problem is.  Matt posted the other day 'it looks like it was caused by a low down feature in Apache which has been disabled for now'.  Well, sorry, Matt.  That feature may be disabled, but the problem is still there (as it has been on and off since October last year).  A while ago, a new problem was created (as Graham posted) by a change intended to correct a different problem.  No evidence of testing there either.

I understand why the NBN corporate body is reluctant to admit what is obvious to the rest of us - namely that Gateway 5 is a complete mess which will not be sorted by tinkering around changing and disabling features here and there in the hope that the right feature has been targeted and that there are no unwanted knock-on effects.  I think any corporate embarrassment will be short-lived if there is an acceptance that G5 is a disaster, and needs a proper analysis of the whole system, right down at the lowest levels of the software structure and design, to find out what is not working as intended and why.  And then sorted.

Murdo

Syrphus's Website

2 Reply by Syrphus

  • Syrphus
  • Veteran
  • Offline
  • From: Strathpeffer, Highland
  • Registered: 16-12-2008
  • Posts: 535

Re: Yes, you've guessed it.

And the same again today 0742h, but accessing G4 gives a screed of Apache errors.

This after apparently working fine for much of yesterday (Sunday).

This demonstrates once more what has been evident for the past 6 months - this resource is based on an unstable design and you can never be sure whether it will work or not, or why.

The related Opera compatibility with the IMT is an example.  From being unpredictable in behaviour, it became impossible and then recently (when the database allows at all) G5 IMT works fine in Opera.

This apparently without any targeted intervention at the NBN end.  So I assume that in twiddling with some other problem, you have inadvertently solved the Opera compatibility.  Great - as long as more twiddling does not send us in reverse.

M.

Syrphus's Website

3 Reply by matt.debont (edited by matt.debont 28-04-2014 08:24:41)

  • matt.debont
  • Veteran
  • Offline
  • From: Peterborough
  • Registered: 04-11-2013
  • Posts: 145

Re: Yes, you've guessed it.

We have changed very little of the code within the IMT recently and certainly nothing that would cause issue with Opera (an issue which we have been unable to recreate as of yet, but we are not actively supporting Opera as it is), most of our efforts are going into trying to figure out what exactly the root cause of the mapping instability actually is. I thought I had found the issue which was related to an issue with Apache using a Winsock API which occasionally has issues with anti-virus and windows servers in general, not really an issue we could have predicted.

Solving this issue has uncovered another issue, which I will not bore you with the details of, but the upshot of all these issues is that we have found no way to trigger these issues manually, they seem to occur at random, and as such they are exceedingly difficult to track down without being able to force the issue to occur.

Once again however we apologise for any issues this may have caused and we are working towards a permanent solution regarding the mapping stability, however we have recently greatly improved the stability of the API and the databases backing the gateway and now mapping stability is one of our current top priorities.

Matt Debont
Application Developer
Joint Nature Conservation Committee, Monkstone House, City Road, Peterborough PE1 1JY, UK

4 Reply by Syrphus

  • Syrphus
  • Veteran
  • Offline
  • From: Strathpeffer, Highland
  • Registered: 16-12-2008
  • Posts: 535

Re: Yes, you've guessed it.

'they seem to occur at random'

I think that is another way of saying 'this resource is based on an unstable design', so we agree at least on that.

Re the Opera thing, currently IMT seems to work OK when the Gateway allows any mapping to work.  That was not the case before, on the same Opera version and platform, so clearly something has changed at your end, whether deliberate or otherwise.

M.

Syrphus's Website