2010-07-29T01:29:26  <okibot> New news from nightlybuilds: Opticks4.3.X-Nightly-Windows #100 (SUCCESS)
2010-07-29T05:09:01  *** goffena has joined #opticks
2010-07-29T05:23:53  *** kstreith has quit IRC
2010-07-29T05:24:02  *** kstreith has joined #opticks
2010-07-29T05:24:02  *** ChanServ sets mode: +o kstreith
2010-07-29T05:48:06  *** dsulgrov has joined #opticks
2010-07-29T05:54:47  *** goffena has quit IRC
2010-07-29T05:56:26  *** goffena has joined #opticks
2010-07-29T05:59:37  *** mconsidi has quit IRC
2010-07-29T06:23:56  *** tclarke has joined #opticks
2010-07-29T06:23:56  *** ChanServ sets mode: +o tclarke
2010-07-29T08:48:18  *** rforehan has joined #opticks
2010-07-29T09:44:11  <goffena> .
2010-07-29T09:44:11  <goffena> .
2010-07-29T09:44:26  <goffena> Are DataAccessors and Pagers threadsafe?  I suspect no.
2010-07-29T10:59:25  *** gmartin_cn has joined #opticks
2010-07-29T12:03:53  <kstreith> DataAccessors no
2010-07-29T12:03:56  <kstreith> Pagers, yes, they are required to be
2010-07-29T12:04:03  <kstreith> if a Pager implementation is not thread-safe, then it is not properly implementing the interface
2010-07-29T12:06:43  <goffena> Could the following be threadsafe:  two data accessors using datarequests of the same block of data, and 2 worker threads, and a threadsafe pager.
2010-07-29T12:06:52  <kstreith> if there is one DataAccessor per-thread
2010-07-29T12:06:53  <goffena> assume the threads are only reading data via the accessor
2010-07-29T12:07:00  <kstreith> then yes, that is meant to be supported by the design
2010-07-29T12:07:17  <kstreith> the pager is required to return a different RasterPage* instance for each request, even if two requests are to the same underlying block in the file
2010-07-29T12:08:42  <kstreith> this is by design to aid thread safety
2010-07-29T12:11:05  <goffena> ok.
2010-07-29T12:11:05  <goffena> .
2010-07-29T12:11:06  <goffena> Another question
2010-07-29T12:11:07  <goffena> unrelated.
2010-07-29T12:11:25  <goffena> Has anyone ever seen an importer hang on  a large dataset, near completion of the import.
2010-07-29T12:11:32  <goffena> i have a 4000x4000x30 dataset.
2010-07-29T12:12:48  <kstreith> i believe that is the start-up time of the SpatialDataView and RasterLayer, etc.
2010-07-29T12:12:51  <goffena> The progress dialog was stating 99% at the hang
2010-07-29T12:12:54  <kstreith> we don't report progress in that section of code
2010-07-29T12:13:10  <kstreith> do you see progress in the status bar (bottom left)?
2010-07-29T12:13:11  <goffena> when i debugging it appeared to hang immediately after a call to update progress dialog
2010-07-29T12:13:13  <kstreith> calculating statistics, generate imge?
2010-07-29T12:13:24  <goffena> i've never noticed progress in status bar, maybe i'll have to look sometime.
2010-07-29T12:13:30  <goffena> Computing statistics of band 1: 100%
2010-07-29T12:13:31  <kstreith> it just has a percentage number
2010-07-29T12:13:33  <goffena> i subsetted 3 bands
2010-07-29T12:13:39  <goffena> ok.  didn't know about status bar progress percent.
2010-07-29T12:13:45  <goffena> hdf5 dataset
2010-07-29T12:15:00  <kstreith> how long is long?
2010-07-29T12:15:04  <kstreith> for the hang?
2010-07-29T12:15:04  <goffena> 5 minutes
2010-07-29T12:15:09  <goffena> i suspect will turn into forever
2010-07-29T12:15:16  <goffena> i do only have 256MB video card
2010-07-29T12:15:26  <kstreith> haven't seen that myself
2010-07-29T12:15:27  <goffena> i would think that would be fine even if statistics were calculated on the video card somehow
2010-07-29T12:15:29  <kstreith> i've seen delays, like 1 minute or less
2010-07-29T12:15:33  <goffena> i had commented out one call to progess dialog
2010-07-29T12:15:35  <goffena> then it would go further
2010-07-29T12:15:37  <kstreith> what interleave?
2010-07-29T12:15:41  <kstreith> what processing location?
2010-07-29T12:15:50  <goffena> georeferencing would complete, and then hang right after another call to update progress dialog
2010-07-29T12:15:50  <goffena> bsq
2010-07-29T12:15:54  <goffena> Opticks 4.4.0rc1
2010-07-29T12:16:08  <goffena> todd mentioned if progress dialog was updated within an openmp pragma, something like this could happen
2010-07-29T12:16:18  <kstreith> true, progress must be reported on main thread
2010-07-29T12:16:19  <goffena> but i turned openmp support off for the importer's plugin module and built
2010-07-29T12:16:21  <goffena> still had same problem
2010-07-29T12:16:30  <kstreith> i'm using the image you gave me for white tiles using 4.4.0 (3840x3840x50)
2010-07-29T12:16:31  <goffena> weird thing is, i can load a 950x950 dataset from the same file without problem
2010-07-29T12:16:33  <kstreith> and don't have any problem like this
2010-07-29T12:16:47  <goffena> good idea.  let me try that on this machine too
2010-07-29T12:23:29  <goffena> uuggh.  Opticks will not recognize the test file kstreith referenced
2010-07-29T12:23:35  <goffena> regardless it worked on kstreith's machine
2010-07-29T12:23:41  <goffena> must be something with my importer or georeferenceing plugin, i guess
2010-07-29T12:23:48  <kstreith> it's an ice file
2010-07-29T12:24:18  <goffena> i meant my original problem must be due to my importer or georef plugin
2010-07-29T12:25:41  <kstreith> ah, i see
2010-07-29T12:25:42  <goffena> i'm pretty sure both ice and generic hdf5 importers refused to recognize the white tile dataset
2010-07-29T12:25:49  <kstreith> ah, you mean the original one
2010-07-29T12:25:54  <kstreith> i meant the one you gave me that was generated using cube creator
2010-07-29T12:25:58  <kstreith> that also exhibits the problem
2010-07-29T12:26:03  <kstreith> white tile problem
2010-07-29T12:26:18  <goffena> yes checked again.  neither recognize the white tile dataset
2010-07-29T12:28:27  <goffena> attempted to load a 2000x2000 dataset from the original file.
2010-07-29T12:29:56  <goffena> so far 90 seconds and counting, at 100%
2010-07-29T12:30:06  <kstreith> stepping out for a bit
2010-07-29T12:30:07  <goffena> the datasets that don't load are parents of smaller datasets that do load
2010-07-29T12:31:20  <goffena> i am stepping out as well
2010-07-29T12:57:29  *** kstreith has quit IRC
2010-07-29T12:58:59  *** goffena has quit IRC
2010-07-29T12:59:00  *** tclarke has quit IRC
2010-07-29T12:59:20  *** dsulgrov has quit IRC
2010-07-29T18:45:25  *** rforehan has quit IRC
2010-07-29T19:17:27  *** gmartin_cn has quit IRC