[RAS] barro bug

[Thomas Krichel]

  Thomas Krichel writes
> 
> 
>   The problem with Barro is still with us.
> 
> Mon Jul 13 23:11:37 2009  request:
>  source: /home/aras/acis/bin/updareq [17370]
>  collection: ACIS
>  update: /r/b/rbarro at harvard.edu.xml ()
> U DATAFILE_START: r/b/rbarro at harvard.edu.xml
> RECORD OLD: rbarro at harvard.edu
> Mon Jul 13 23:11:49 2009 Error: cannot store record 'repec:per:2005-04-12:robert_j_barro' in ObjectDB at /home/aras/acis/lib/ARDB.pm line 190.
> 
>   Barro and Heckman (the other case) are large
> 
> aras at nebka:~$ ls -l acis/userdata/r/b/rbarro\@harvard.edu.xml 
> -rw-r--r-- 1 aras aras 6165312 2009-07-10 16:53 acis/userdata/r/b/rbarro at harvard.edu.xml
> 
>   I suspect there is some size limitation that we are hitting.
> 

  I don't think there is any limitation built in into ACIS, but 
  I read

http://www.mysql.com/news-and-events/newsletter/2003-09/a0000000237.html

The theoretical limit in MySQL 4.0 is 2G, however each blob requires generally to have 3 copies of it in the memory (stored in various buffers) so you need a lot of memory, if you have large BLOBs stored in MySQL. This is the reason, why the theoretical limit can be reached only on 64bit systems. The Practical limits are around some hundreds of megs per BLOB.

  We need more memory, or we can just ignore this issue as the
  blobs are not actually used?


  Cheers,

  Thomas Krichel                    http://openlib.org/home/krichel
                                RePEc:per:1965-06-05:thomas_krichel
                                               skype: thomaskrichel