[daip] "Bring out the canons"
Eric Greisen
egreisen at nrao.edu
Thu May 16 13:21:50 EDT 2013
R. Craig Walker wrote:
> On my Mac laptop (Macbook Pro - Intel) with an AIPS that was updated
> within the week, I tried to MOVE some files from one user to another
> in preparation to move them back to a different (external) disk under the
> original user (originally guessed to be an easier way to do bulk transfers
> than sending to a new name, then naming back). Most files went ok. But
> the IBM002, IBM003 etc files from the multi-resolution clean got copied
> ok, but the original user catalog shows that they also still exist.
> During the operation of MOVE, there were complaints about the HI table not
> being found. The files in the original user have a READ flag. I tried a
> RECAT to make it easier to do loop operations, but it claimed no files
> needed to be renumbered, which doesn't seem right - there are big gaps in
> the sequence. When I try to look at those files (IMH), it doesn't work -
> FILE MISSING. When I try a ZAP, I get HEADER FILE MISSING OR EMPTY -
> BRING OUT THE CANONS. But the file does go away. I'll speculate that the
> missing HI table somehow caused MOVE to remove the files from the disk
> (actually just change the names), but not remove them from the catalog.
>
> Next I try to move them back, but this time a real copy and delete is
> needed because the disk changes.
>
> That seems to have worked, after some confusion with disk numbers. But
> the higher number beams (lower resolution) again have READ flags on the
> source disk. There were complaints again about the HI file not found.
>
> Is there a better way to move files from one disk to another without
> jumping through hoops like this? When working with external drives, it is
> an occasional need, especially when the sequence number really means
> something (for me, the imaging loop iteration) and needs to be preserved.
> I know a copy can't work because of AIP's reluctance to have two files of
> the same name. But a move should be possible with some renaming in the
> middle without having to do it by hand.
Move treated the missing HI as an error and quit abruptly - all other
errors were handled with great care. Tomorrow's MNJ will have a change
to more or less ignore the HI error and finish normally.
Thanks
Eric
More information about the Daip
mailing list