Date: Wed, 5 Jun 2013 13:29:37 -0400
From: "Blundell, Raymond" 
To: "Marrone, Daniel P - (dmarrone)" 
Cc: Jun-Hui Zhao , Glen Petitpas , Ken Young ,
     "cqi@cfa.harvard.edu" 
Subject: Re: track with crate loss
Parts/Attachments:
   1   OK     73 lines  Text (charset: ISO-8859-1)
   2 Shown   ~74 lines  Text (charset: ISO-8859-1)
----------------------------------------

Hi Dan

Sorry about that, in DC right now.� Let me know what you decide to do.� Interesting that you have difficulties 20% of the
time.� Will meet with Taco tomorrow for opinion.

Regards,
Ray




On Wed, Jun 5, 2013 at 1:18 PM, Marrone, Daniel P - (dmarrone)  wrote:
      Hello all,

      In my latest track, 130524_03:24:11, a fuse blew in crate 10 and so crates 4/10 were removed from the project.
      Unfortunately, the data continued to be written to the same file. I can't load this file in miriad because the
      data structure changes during the track. MIR also has some trouble and has to revert to "OLD READDATA" to read
      it, it's crunching away on that now and probably will be for hours.

      I can see a few ways forward, all of which would be useful in many applications (at least for me, since I seem to
      have trouble loading about 1 out of every 5 tracks that I get). I email to solicit as much help as I can get.

      Jun-hui, it's possible that smalod could read this if the nscans input worked for 4GHz data. I can find the
      boundary scan in MIR and could read only before or after it with this keyword. Do you think this would work, and
      if so, is there any chance you could implement nscans for 4GHz data? If this wouldn't work, could you not add a
      little error checking to the data reading so that it would report a change in the data structure and exit
      gracefully with the data it can access?

      Taco, one time a long time ago you rewrote a raw data file for me when it was corrupted. I've asked you before
      about making it possible for me to hack at the code that could do this but not gotten a response. I could take
      that as a hint that it's impossible, but I am not good at subtlety. In this case (and probably most cases, since
      I could always leave the bad data out of the file), just code that could divide a data file into parts at a
      certain scan number would be adequate. Can anything be done on this front?

      Charlie, I could load this through mir and write out to miriad. However, the idl2miriad writer does not fill the
      miriad header with enough variables to do everything I need. I could come up with specific examples if you are
      willing to update that code to make it more closely resemble native miriad files.

 glen, I don't know that errors like this happen often, but starting a new file might be worth suggesting to the
      operators if there is some catastrophic change during a track.

      Thanks,
      Dan


      Dan Marrone
      Assistant Professor
      University of Arizona
      Department of Astronomy/Steward Observatory
      933 North Cherry Avenue
      Room N314
      Tucson, AZ 85721

      Office: (520) 621-5175
      Fax: (520) 621-1532