Thread Rating:
  • 0 Votes - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
github commit to fix bug #74
23-12-2011, 04:21 PM
Post: #1
github commit to fix bug #74
There's a recent commit on github entitled "Temporarily disable sending Removed/Added events when a device moves location. See #74". Where can I find details of bug #74? I'd like to understand what problem this commit is solving so that I can decide whether I should be using the version of the code without this commit, or with it.

Many thanks, Simon.
Find all posts by this user
04-01-2012, 02:04 PM
Post: #2
RE: github commit to fix bug #74
(23-12-2011 04:21 PM)simoncn Wrote:  There's a recent commit on github entitled "Temporarily disable sending Removed/Added events when a device moves location. See #74". Where can I find details of bug #74? I'd like to understand what problem this commit is solving so that I can decide whether I should be using the version of the code without this commit, or with it.

Many thanks, Simon.

Our bug database isn't externally visible. We'd like to change this but its not our highest priority. You're not missing much in this case anyway - the report only says

Device list should infer a Removed if a UPnP device moves location.
The code is in place for this but its making tests unstable so has been disabled until I get time to properly debug it.

The problem tests are the ones that iterate through all devices offering the ConnectionManager service. On our network with ~70 such devices, many of which are frequently restarted, we occasionally see device references being leaked. This can cause our nightly tests to fail which blocks pushing of updates to github and can block progress in other downstream projects. These problems only appeared when I added checks to catch devices which silently shutdown then reappear on a new location. I disabled this test and the problems were immediately solved.

I hope to investigate and fix this propery in the next couple of weeks. Nothing bad will happen if you choose to fork ohNet to retain the checks for moved devices. That said, I personally wouldn't bother - the chances of finding that devices move location like this are very rare in real use and the costs of maintaining a fork probably outweighs the benefits you'll get.

I'll let you know when bug #74 is fixed.
Find all posts by this user
04-01-2012, 10:52 PM
Post: #3
RE: github commit to fix bug #74
Hi Simon,
Happy New Year! I hope you had a good break over Christmas and New Year.

Thanks for the explanation. I'll go with your recommendation to use the currrent version without checks for moved devices.

On a completely different subject, could you have a word with the forum administrator about removing/blocking the large number of spam postings on the General forum? It's very annoying to be receiving such a large number of "new thread" notification emails for these.

Simon
Find all posts by this user
06-01-2012, 11:12 AM
Post: #4
RE: github commit to fix bug #74
Bug #74 is now fixed on github.

The spam messages are gone (again) and the bot responsible for recent posts has been blocked. This should at least reduce the amount of spam we see.
Find all posts by this user
06-01-2012, 11:38 AM
Post: #5
RE: github commit to fix bug #74
(06-01-2012 11:12 AM)simonc Wrote:  Bug #74 is now fixed on github.
Thanks. I'll give this a try.

Quote:The spam messages are gone (again) and the bot responsible for recent posts has been blocked. This should at least reduce the amount of spam we see.
Thanks. To reduce future occurrences, it might help to change the signup question to one that can't easily be answered by a bot. It would be very easy to write bot code that correctly answers the question "How many letters are there in the word xxxx?" for all cases of xxxx.
Find all posts by this user


Forum Jump: