Changes between Initial Version and Version 1 of Ticket #4330, comment 5


Ignore:
Timestamp:
Mar 9, 2019, 9:53:35 AM (5 years ago)
Author:
hranalli

Legend:

Unmodified
Added
Removed
Modified

  • Ticket #4330, comment 5

    initial v1  
    1 It's been a while since I made this patch, as I didn't want to submit it until after several test runs, but I am pretty sure the CREATE TABLE statement also processes the table data. To figure out the problem, I added logging to see where it crashed. And it was in the  middle of the CREATE TABLE loop that it ran out of memory. The change I made there resolves the problem. I've run the script several times on our database, and it always fails without the patch, and succeeds with it.
     1Edited as I just looked more closely at my patch...
    22
    3 I adjusted the COMMENT area because, as there was no intermediate processing required, it seemed reasonable to remove a similar, potential failure point (though I agree an unlikely one, given the data sizes involved). Other sections which buffer data need the buffer to make decisions, and so would be harder to adjust.
     3It's been a while since I made this patch, as I didn't want to submit it until after several test runs, but it is the INPUT statement that processes the table data. To figure out the problem, I added logging to see where it crashed. And it was in the  middle of the INPUT loop that it ran out of memory. The change I made there resolves the problem. I've run the script several times on our database, and it always fails without the patch, and succeeds with it.
    44
    5 You can certainly leave out the patch on the COMMENTS section. I do know the patch on CREATE TABLE solves the problem. There may be a more fundamental issue, but I'm not a Perl programmer or PostGIS expert.
     5I adjusted the CREATE TABLE and COMMENT areas because, as there was no intermediate processing required, it seemed reasonable to remove a similar, potential failure points (though I agree unlikely ones, given the data sizes involved). Other sections which buffer data need the buffer to make decisions, and so would be harder to adjust.
     6
     7You can certainly leave out the patch on those sections. I do know the patch on INPUT solves the problem. There may be a more fundamental issue, but I'm not a Perl programmer or PostGIS expert.
    68
    79