PDA

View Full Version : Portal loosing .py file and more


martinnossel
11-30-2008, 10:04 PM
I have now had the current file I am working on disappear twice and I mean the file is cleared on the hard drive. The is the xxx.py in the directory and the size is 0.

One was on my machine. And once on the test site machine while connected via gotomypc.

Both times I did not have a backup as I am trying to debug the programs and its work in progress, and is more than infuriating. It happens as you click the tab to bring up the code.

Also the rename function does not work in updating the nodeX name in portal, its cleared after a node reboot or “network..new “configuration. I have had to use the node.name = “loadNvParam(8)”_sent_by_node (every time there is communication from a node to portal)

I have a test system set up that been running for about 10 days. In the test there are 9 nodes. At the moment the system has been running 10 days each node sending data ever 5 seconds. No sleep.

I want them to sleep for varying times from 5sec to 15min depending on whatever.

The set up is star. (I think, as all nodes are in range of the bridge (not sure how to confirm this but would like to be able to.))


I wrote and tested software with only one node that allows the node to wake up, send data and check if it can go back to sleep. The bridge is always awake. It looked very nice with the ammeter pulsing for less than a second and the data appearing in the portal chart.


I just loaded one of the test nodes and the ‘portal_node’ with above software. One node sleeping and waking up for a instant to transmit every 5sec, the other 8 transmitting every 5 seconds but not sleeping.(the awake nodes send 4 short strings 2 to the chart, and 2 to the log). The sleeping node send 10 variables (displayed in portal), the above 4 strings and requests a ‘can I sleep’ variable. The system immediately bogged down completely some data was received from 2 nodes.

I am about to change the NV parameters to ‘listen out’ and reduce the ‘discovering calls’.

Please can I get some details on what portal is doing in the background so I to optimize this system.

Even with just a one node system waking up and sleeping, if I press the “watch nodes button” the poor event log goes berserk, with 3sec discovery calls etc.

I see in the new sleepy mesh examples, (thanks, they were informative, we need more of the same) that in one you poll the sleepy mesh (for the above reason?). If this is the case I am going to get more speed out of a bunch of couriers with ‘post its’ than a mesh with 600 or more nodes and will need car batteries to power the nodes as they stay awake waiting there chance to transmit. How I am supposed to divide up the network and debug such a system after what I am struggling with now, makes me want to get a job as a street sweeper. (which I will soon probably end up having to do)


Martin

martinnossel
11-30-2008, 11:39 PM
Just happend again, when I clicked the save button. what a PITA. I did have a back up this time, but its still 10 lines of painfull code to redo.

You have got to do somthing about this!!!

I reloaded portal. Its been open for days logging, but isnt that the intention?

Martin

kbanks
12-01-2008, 09:46 AM
The set up is star. (I think, as all nodes are in range of the bridge (not sure how to confirm this but would like to be able to.))

1) If your network is in fact "star" (by virtue of physical node placement), then you don't need to use the "sleepy mesh" approach.

2) To confirm all of your nodes are "one hop" from the bridge (which means they must be two hops from Portal), go to Preferences and change your "Ping TTL" to 2, and verify you can still talk to all of your nodes.

kbanks
12-01-2008, 09:51 AM
Even with just a one node system waking up and sleeping, if I press the “watch nodes button” the poor event log goes berserk, with 3sec discovery calls etc.

It says on page 20 of the newest "Sleepy Mesh" app note to

Make sure the Watch Nodes function is off. This function is not aware of the Sleepy Mesh, and will slowly fill your Portal queue with extra "pings".

Why are you pressing the "watch nodes" button?

kbanks
12-01-2008, 09:55 AM
Just happend again, when I clicked the save button. what a PITA. I did have a back up this time, but its still 10 lines of painfull code to redo.

You have got to do somthing about this!!!

I reloaded portal. Its been open for days logging, but isnt that the intention?

I've no idea what is different about your setup versus mine. I run Portal for weeks at a time, and do all of my script editing with it.

Portal does support interoperation with external editors, you might try that (using something like notepad, etc.). Then if something happened to the file on disk it would still be in the other (external) editors memory.

mgenti
12-01-2008, 09:57 AM
I have now had the current file I am working on disappear twice and I mean the file is cleared on the hard drive. The is the xxx.py in the directory and the size is 0.

I am sorry to hear that you are loosing the file that you are working on. This is not an issue we have seen yet, so I would like to get a few more details on when this is occurring.

Is Portal the only application that has the file open? For example, do you have this file open in Portal and Notepad at the same time? If this file is open in another program, are you making changes to the file in both programs?
Does this happen when you click the save button in Portal's editor or when you switch to the editor tab after being somewhere else in Portal?

cbowen12345
03-06-2009, 08:45 PM
I am suddenly having the same problem as Martin. With Portal started but no bridge attached to the computer, I will open a new script file using the Portal editor and then use "save as" to establish the file name. After typing code or copying it from another saved py file I attempt to save a copy by touching the disk icon and WAAM!!!! all the code I had entered is gone. The file is saved with the correct name but there is nothing in the file. The circle that indicates that portal is processing code shows up at the tip of the pointer and will not go away until I close portal (2.1.22). Even more frustrating is that it does not happen every time. I keep making copies of my code because I don't know when it will the next "Save failure" event will occur. Any suggestions?

Only Portal is running at the time the problem occurs

Thanks

mgenti
03-09-2009, 09:46 AM
Is this something that you can reliably reproduce? If so, would you be willing to try a version of Portal that could provide us with some extra information to help determine the root cause of this problem?

cbowen12345
03-09-2009, 02:02 PM
Thanks Mark,

When I get finished with my day job I will try to reproduce the sequence of events/keystrokes that kept resulting in the problem. I will let you know.