#850 Flush needs to be supported for recorders too,

[nikhilgupta10]

It seems flush is only supported for collectors and not recorders. I don't know why but it seems pretty obvious that recorders need it too. See ticket 848 for a case in point.,

[nikhilgupta10]

nikhilgupta10 imported these comments from Sourceforge: The user dchassin does not exist anymore. Therefore assigning this to afisher1. "dchassin":Apparently collectors did not support flush either. It is group_recorder that does. Not sure about about group_collector though.,

"dchassin":The fix is completed and the value \int32 flush\ has been added to ecorder\ and \collector. The default lush=-1\ means flush when the buffer is full. The value lush=0\ mean flush each time a line is written. The value lush>0\ means flush every 'flush' simulation seconds (i.e., when lush mod clock == 0).

I considered implementing flush<-1 meaning lush ever '-flush-1' seconds of real time, but that seemed a bit complicated to understand and perhaps too much overloading. If this is really desired, then we need two parameters lush_mode\ and lush_interval\ instead. Pass.

Ready for validation.,

"dchassin":- assigned_to: Jason Fuller

• Resolution: none --> validate
• Component: none --> tape
• Priority: none --> major
• Version: --> trunk

• Type: defect --> enhancement ,

  "dchassin":The new flush property is documented in recorder and collector on the wiki pages (see http://gridlabd.me.uvic.ca/gridlabd/wiki/index.php/Recorder and http://gridlabd.me.uvic.ca/gridlabd/wiki/index.php/Collector). A simple implementation is done in $SVN/ticket/850 and ready for validation.

Note that this is very difficult to test automatically because the behavior is to change when data becomes available in the output file, rather than whether it becomes available. No autotests have been added to verify that this works correctly on all platforms. We will have to rely on user experience to verify the functionality.,

"jcfuller":- status: new --> assigned

• assigned_to: Jason Fuller --> Matt Engels
• Milestone: Unscheduled --> Version 4.0 RC1

• Priority: major --> critical ,

  "jcfuller":Not sure how to validate this either. Matt, let's discuss before you get going on this.,

  "dchassin":I had a thought about validation. You could probably use the #script macro to check for a non-zero file size after postcommit. If flush works the test would succeed. If flush fails to operate the file would still be zero size and the test would fail.,

  "dchassin":- assigned_to: Matt Engels --> David P. Chassin ,

  "dchassin":This needs to be integrated into trunk.,

  "dchassin":- Milestone: Version 4.0 RC1 --> Knothole Interim ,

  "andyfisher":after integration can this be closed?,

  "dchassin":Yes,

  "andyfisher":- status: assigned --> closed ,

  "andyfisher":Merged into trunk at r5508. Closing ticket.,