Re: [Gusdev-gusdev] everything works except commit!

[MICHAEL L. <lu...@cs...>]

Steve-
I will try this.  But some points to consider:
1.The previous data was entered using a plugin called
LoadBlastSimilaritiesPK.pm, included in the GUS release.
2.The "sequences" increments each time regardless of whether commit is
turned on.   Maybe not elegant, but I guess this doesn't matter,huh?

Thanks,

Michael Luchtan
http://www.cs.uga.edu/~luchtan


On Mon, 19 Apr 2004, Steve Fischer wrote:

> Michael-
>
> let me try to cobble together an answer.   first, i am the culprit
> responsible for that plugin.   but, i think it should be ok.
>
> about "sequences."    that term refers a database sequence, not a
> biological one.   it is a facility supported by oracle (and postgres)
> that provides an incremented number.    the plugins and object layer use
> such sequences to get the next primary key when inserting a row.
>
> so, i suspect that your initial data in those tables was added "outside"
> the plugin structure.
>
> i think the thing to do here is:
>   1. avoid adding data without using either the plugins/object layer or
> using the sequence directly
>   2. to correct the out-of-sync sequence, just run it in a loop
> advancing it past 3000 or whatever it needs to be.
>   3. re-instate the code that uses the sequence
>
> I would try that first, and then lets see if you still get the
> constraint errors.
>
> steve
>
> MICHAEL LUCHTAN wrote:
>
> >Hello everyone-
> >I have been attempting to load blast similarities into GUS with the plugin
> >LoadBlastSimFast.pm after already loading some data in with the plugin
> >LoadBlastSimilarityPK.pm (which was slow compared with the Fasst version).
> >Well, the first thing that LoadBlastSimFast.pm does is start complaining
> >about PrimaryKey uniqueness constraints being violated.  It used some
> >fancy Oracle SQL that I didn't understand:
> >select dots.similarity_SQ.NEXTVAL from DUAL
> >Which by debugging I found out started at one, and increased each time we
> >tested the plugin, regardless of commit not being turned on.  Naturally this crashed, because
> >we already had about 3,000 entries in the dots.similarity table and
> >we already had a sequence_id with value 1.  There
> >was likewise a problem with the dots.similarityspan table.
> >
> >Since I only having a workingman's knowledge of OracleSQL, I switched it
> >so that it starts with
> >select count(*) from dots.similarity
> >And increases this by one for each entry.  Likewise for
> >dots.similarityspan.  This seemed to fix the problem, and the plugin works
> >great on each file tested, claims to have entered all the correct data,
> >and exits gracefully.  Yet when I run the same command line with --commit
> >appended to the end, we start getting these Primary Key constraints
> >violated errors again.
> >
> >Perhaps I do not understand what happens with commit.  I thought that
> >these kind of errors would be caught with a regular run without commit
> >turned on.
> >Any help?
> >
> >
> >
> >
> >Michael Luchtan
> >http://www.cs.uga.edu/~luchtan
> >
> >
> >
> >
> >
> >-------------------------------------------------------
> >This SF.Net email is sponsored by: IBM Linux Tutorials
> >Free Linux tutorial presented by Daniel Robbins, President and CEO of
> >GenToo technologies. Learn everything from fundamentals to system
> >administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
> >_______________________________________________
> >Gusdev-gusdev mailing list
> >Gus...@li...
> >https://lists.sourceforge.net/lists/listinfo/gusdev-gusdev
> >
> >
>
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by: IBM Linux Tutorials
> Free Linux tutorial presented by Daniel Robbins, President and CEO of
> GenToo technologies. Learn everything from fundamentals to system
> administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
> _______________________________________________
> Gusdev-gusdev mailing list
> Gus...@li...
> https://lists.sourceforge.net/lists/listinfo/gusdev-gusdev
>