2010-06-14T04:19:49  *** goffena has joined #opticks
2010-06-14T04:25:11  *** goffena has quit IRC
2010-06-14T04:28:01  *** goffena has joined #opticks
2010-06-14T04:35:46  *** yuguess has joined #opticks
2010-06-14T05:14:14  *** tclarke has joined #opticks
2010-06-14T05:14:15  *** ChanServ sets mode: +o tclarke
2010-06-14T05:35:26  *** yuguess has quit IRC
2010-06-14T05:39:27  *** yuguess has joined #opticks
2010-06-14T05:39:40  <yuguess> hello
2010-06-14T05:40:17  *** kstreith has joined #opticks
2010-06-14T05:40:17  *** ChanServ sets mode: +o kstreith
2010-06-14T05:41:59  <yuguess> Hi
2010-06-14T05:42:05  <tclarke> hello0
2010-06-14T05:42:06  <kstreith> hi
2010-06-14T05:42:26  <yuguess> May I know who write spectral part of the code ?
2010-06-14T05:43:51  <yuguess> there seems a small error in the code
2010-06-14T05:44:01  <kstreith> it's been a team, hasn't been just one person
2010-06-14T05:44:03  <kstreith> which plug-in?
2010-06-14T05:44:06  <yuguess> that mnf.cpp
2010-06-14T05:44:18  <kstreith> can you reference a function or line #?
2010-06-14T05:45:53  <yuguess> en...Line 2161
2010-06-14T05:46:02  <yuguess> pMatrix[band2][band1] /= pixCount2 - 1
2010-06-14T05:46:10  <yuguess> pMatrix[band2][band1] /= (pixCount2 - 1) would be better
2010-06-14T05:46:16  <tclarke> they are equivalent
2010-06-14T05:46:21  <tclarke> since the - has higher precendence than the /=
2010-06-14T05:46:49  <kstreith> agree with tclarke, both lines are equivalent
2010-06-14T05:48:21  <tclarke> it would be different if the line were written as: pMatrix[band2][band1] = pMatrix[band2][band1] / (pixCount2 - 1)    in which case the parens are required
2010-06-14T05:48:25  <yuguess> oh...it is, sorry
2010-06-14T05:48:35  <kstreith> np
2010-06-14T05:48:35  <tclarke> no problem...we'de rather you mention potential problems when you see them
2010-06-14T05:48:38  <tclarke> as there are definately bugs in the software
2010-06-14T07:15:21  *** sbmiller has joined #opticks
2010-06-14T07:23:03  *** dsulgrov has joined #opticks
2010-06-14T07:24:47  *** yuguess has quit IRC
2010-06-14T07:48:10  *** gmartin_cn has joined #opticks
2010-06-14T10:54:00  <tclarke> the meeting will begin in a few minutes
2010-06-14T11:01:25  <tclarke> time to begine the meeting with attendance
2010-06-14T11:01:26  <tclarke> here
2010-06-14T11:01:33  *** dadkins has joined #opticks
2010-06-14T11:01:37  <gmartin_cn> here
2010-06-14T11:01:37  <dadkins> here
2010-06-14T11:01:38  <sbmiller> here
2010-06-14T11:01:44  <goffena> here
2010-06-14T11:01:49  <dsulgrov> here
2010-06-14T11:01:49  <mconsidi> here
2010-06-14T11:01:51  <tclarke>    1.  Attendance
2010-06-14T11:01:51  <tclarke>    2. Announce agenda
2010-06-14T11:01:51  <tclarke>    3. Approve May 17, 2010 Meeting Notes. Ask for review and address any problems seen in the notes.
2010-06-14T11:01:51  <tclarke>    4. Introduce sbmiller
2010-06-14T11:01:51  <tclarke>    5. Google Summer of Code updates
2010-06-14T11:01:51  <tclarke>    6. Open Questions
2010-06-14T11:01:51  <tclarke>    7. Summary and closing
2010-06-14T11:01:51  <kstreith> here
2010-06-14T11:01:56  <tclarke> notes from the last meeting are here http://opticks.org/confluence/display/opticksChat/May+17%2C+2010+Meeting+Notes
2010-06-14T11:02:01  <tclarke> if you see any errors please fix them or let me know
2010-06-14T11:02:08  <tclarke> if nothing has been brought up by the end of the meeting, I'll assume they are accepted
2010-06-14T11:02:15  <tclarke> we'll start by introducing a new developer, sbmiller
2010-06-14T11:02:22  <tclarke> could you briefly introduce yourself?
2010-06-14T11:03:52  <sbmiller> sure!  I'm an intern here at Ball for the next ten weeks, working on various projects for opticks.
2010-06-14T11:04:02  <sbmiller> currently, I'm assigned to take a look at improving performance for the MODIS importer and am working with mconsidi as my mentor
2010-06-14T11:04:19  <tclarke> for those who don't know, we are working on releasing a MODIS data importer as an open source plug-in
2010-06-14T11:04:25  <tclarke> any questions or comments for sbmiller?
2010-06-14T11:04:27  <tclarke> if not, we'll move on
2010-06-14T11:04:35  <tclarke> ok, google summer of code update
2010-06-14T11:05:44  <tclarke> our two SoC students have been busy
2010-06-14T11:06:03  <tclarke> updates for each student are available weekly on their wiki home pages
2010-06-14T11:06:08  <tclarke> http://opticks.org/confluence/x/MYFC
2010-06-14T11:06:19  <tclarke> and http://opticks.org/confluence/x/IIFC
2010-06-14T11:06:30  <tclarke> they both have started putting up some code so go check out their updates if you are interested
2010-06-14T11:06:38  <tclarke> they will not necessarily be at these meetings due to time zone differences
2010-06-14T11:07:52  <tclarke> so I'll occasionally summarize their work but there won't be a full report each meeting
2010-06-14T11:07:57  <tclarke> any other news or discussion on SoC?
2010-06-14T11:08:06  <tclarke> ok, open discussion...any topics to talk about>?
2010-06-14T11:08:11  <kstreith> programming contest
2010-06-14T11:08:18  <kstreith> the deadline for submission was June 4th
2010-06-14T11:08:21  <kstreith> we are reviewing entries now
2010-06-14T11:08:27  <kstreith> winner(s) will be announce this Friday, June 18th
2010-06-14T11:08:34  <kstreith> .
2010-06-14T11:10:00  <tclarke> ok, anything else?
2010-06-14T11:10:06  <kstreith> python scripting 1.0.1 release
2010-06-14T11:10:15  <kstreith> the 1.0.1 is code complete, you can grab the daily build
2010-06-14T11:10:15  <kstreith> http://opticks.org/hudson/job/Python1.0.X-Daily-Windows/
2010-06-14T11:10:24  <kstreith> the only remaining blocker for the release is documentation
2010-06-14T11:10:31  <kstreith> .
2010-06-14T11:11:57  <tclarke> do we have an estimated release date for Python 1.0.1 ?
2010-06-14T11:12:09  <kstreith> before July 2nd
2010-06-14T11:12:20  <tclarke> any other topics for today?
2010-06-14T11:12:42  <tclarke> ok, we'll end the meeting...notes and the new agenda will be posted to the mailing list later today
2010-06-14T11:12:46  <tclarke> .
2010-06-14T11:22:59  *** dadkins has quit IRC
2010-06-14T11:40:07  *** mconsidi has quit IRC
2010-06-14T11:44:50  <goffena> .
2010-06-14T11:44:50  <goffena> .
2010-06-14T11:47:24  <goffena> I would like to create an on-disk read-only dataset, which is based off the same file as a on-disk read-only dataset already loaded into Opticks.
2010-06-14T11:47:45  <goffena> I may be willing to actually modify the dataset already loaded into Opticks instead of creating a new dataset.
2010-06-14T11:47:54  <goffena> I see RasterElement provides a non-const getDataDescriptor method.
2010-06-14T11:48:10  <goffena> I see RasterUtilities provides a subsetDataDescriptor method.
2010-06-14T11:50:31  <goffena> Does anyone know if this is likely to cause problems with Opticks?
2010-06-14T11:51:01  <kstreith> do you want the new image to appear to be loaded from the original file?
2010-06-14T11:51:09  <kstreith> or do you want it to be an on-disk image generated from an algorithm?
2010-06-14T11:51:10  <goffena> yes, and the new image to be a subset (by band) of the original file.
2010-06-14T11:52:43  <kstreith> ok, do you want it to basically look like the end-user had done a band subset on import? so the new cube is indistinguisable from that?
2010-06-14T11:53:03  <goffena> yes
2010-06-14T11:53:10  <kstreith> if so, look at RasterElement::createChip()
2010-06-14T11:53:19  <kstreith> it is what the chipping window uses
2010-06-14T11:54:57  <goffena> will RasterElement::createChip produce a on-disk/read-only dataset?
2010-06-14T11:55:19  <goffena> the documentation states: The new cube will match the original RasterElmeent's processing lcoatin (on-disk(not read-only) or in-memory)
2010-06-14T11:55:23  <goffena> which confuses me somewhat.
2010-06-14T11:55:33  <kstreith> it will convert an on-disk read-only to on-disk read-write
2010-06-14T11:55:38  <goffena> i don't won't read-write.
2010-06-14T11:55:46  <kstreith> because it physically copys the data over to the new image
2010-06-14T11:57:16  <goffena> I am trying to optimize our algorithms to work on very large datasets (GBytes), and copying takes a long time.
2010-06-14T11:57:32  <goffena> Generally only 1/8 of bands are removed, so there is a lot of data to copy.
2010-06-14T11:57:40  <kstreith> looking...
2010-06-14T11:57:42  <dsulgrov> why have multiple raster elements if they are all going to be subsets of an original?
2010-06-14T11:57:53  <goffena> Historically we have had multiple raster elements.
2010-06-14T11:59:13  <dsulgrov> could you just use the original raster element with processing dimension vectors instead?
2010-06-14T11:59:19  <kstreith> it looks like you could use RasterElement::copyShallow()
2010-06-14T11:59:21  <goffena> I don't think anyone will have a problem if we modify the on-disk/read-only raster element after it is loaded (essentially doing a band subset post-load)
2010-06-14T11:59:23  <kstreith> which does everything except the data and is pretty quick
2010-06-14T11:59:24  <goffena> , if it saves a lot of time.
2010-06-14T11:59:43  <kstreith> and then you would have to wire up a pager in on-disk read-only mode (which means it's not going to work for *ANY* dataset)
2010-06-14T11:59:49  <kstreith> because pager's don't have a copy function
2010-06-14T12:01:16  <goffena> .
2010-06-14T12:01:19  <goffena> What about my original question?
2010-06-14T12:01:22  <goffena> i have a dataset loaded on-disk/read-only.
2010-06-14T12:01:27  <goffena> call it mpRaster
2010-06-14T12:01:51  <goffena> i then call RasterUtilities::subsetDataDescriptor( mpRaster->getDataDescriptor(), .....)
2010-06-14T12:02:00  <goffena> would that work?
2010-06-14T12:03:16  <goffena> Are there any currently known reasons why that wouldn't work?
2010-06-14T12:03:28  <goffena> If it did work, and was supported by Opticks, then that would seem the simplest solution for me.
2010-06-14T12:03:30  <tclarke> The problem isn't necessarily creating the subsetted RasterDataDescriptor
2010-06-14T12:03:32  <tclarke> but using that to import
2010-06-14T12:03:33  <kstreith> that might break more often than it works
2010-06-14T12:03:43  <kstreith> if the RasterElement is already being displayed in a SDV, I expect things are going to start breaking
2010-06-14T12:03:44  <goffena> what do you mean by "import"
2010-06-14T12:03:47  <tclarke> you'd need to get data descriptors from an ImporterResource for the file then make the necessary changes to match the imported dataset then finish the import
2010-06-14T12:05:37  *** mconsidi has joined #opticks
2010-06-14T12:14:26  <kstreith> goffena and I talked off-line
2010-06-14T12:14:30  <kstreith> here is short summary:
2010-06-14T12:14:36  <kstreith> i suggested he instead try to "import the cube" again with a subset
2010-06-14T12:14:44  <kstreith> this doesn't generally work, because we name the image after the filename
2010-06-14T12:14:49  <kstreith> and modelservices uses this as part of the unique key
2010-06-14T12:16:09  <kstreith> i mentioned if he called getImportDescriptors() and changed the parent pointer
2010-06-14T12:16:17  <kstreith> and then imported he could probably get this method to work
2010-06-14T12:16:31  <kstreith> i also reminded him that currently most of the built-in importers disallow subsetting by band when using a processing location of "on-disk read-only"
2010-06-14T12:16:41  <kstreith> he mentioned he was aware of this but they would probably be using their importers which do support this, so no problem
2010-06-14T12:17:00  <kstreith> that's the complete summary
2010-06-14T12:20:51  <goffena> thanks for the help
2010-06-14T13:12:56  *** tclarke has quit IRC
2010-06-14T14:01:23  *** dsulgrov has quit IRC
2010-06-14T18:59:12  *** yuguess has joined #opticks
2010-06-14T19:31:59  <yuguess> hello
2010-06-14T20:11:28  <okibot> New news from nightlybuilds: Opticks4.4Dev-Nightly-Linux #25 (SUCCESS)
2010-06-14T20:11:40  *** gmartin_cn has quit IRC
2010-06-14T20:34:20  *** yuguess has quit IRC
2010-06-14T23:49:11  <okibot> New news from nightlybuilds: Opticks4.4Dev-Nightly-Solaris #24 (SUCCESS)