Jody,

Sorry to be so silent as you and Gabriel work through the arcsde issues.

I'm not sure what to make of these errors.  It could be that someone has
arcmap open and has that layer locked that way.  However, when the
arcsde plugin dies in the middle of a test (say with one of those
SeExceptions below) there's often buggy code in the test suite that
doesn't return the connection to the (global) pool before the test dies.
In this case you'll have one prior test fail, tie up the connections,
and then you'll get a ton of "timeout waiting for idle connection"
errors.  These latter errors are spurious.


Also, my general hint on ArcSDE debugging/coding is that ArcSDE is
actually fairly well coded and seems to work fairly consistently.  The
java api does most of what it needs to do, and is used internally by
several production ESRI products.  So your comment about "my test suite
is faster than my connection" isn't impossible, but in my view it's
unlikely (although you're much closer to the real info than I am!)

Threading issues (which you are exploring explicitly!) does seem to be
well worked out, as long as you use the API the 'right' way.  Where
'right' means follow documentation that is possibly nonexistent, and
likely old or written using the C version of the API.

All of this said, the documentation and "design" of the jsde API range
from planetarily catastrophic to galactically FUBAR.  It's clearly a c
api that's been ported in fits and starts, and by different people, with
different pay grades and education levels.


I'm very happy about the following things that you and Gabriel have
worked out, by the way:

* "Command Queue-ing" a la HibernateTemplate
* An "sde session" ide (a la hibernate)


I'm a little nervous, but holding my tongue about the following things
that you've worked out:

* re-writing the entire sde-connection-facing portion of the API (which
will touch almost all classes)
* any mention of 'abstracting' the jsde api away from the plugin.  This
one in particular makes me really scratch my head...but since it looks
like no-ones trying to take this on, I don't need to warn anyone against
it!


Good luck jody.  Get some caffeine and a big whiteboard and a punching
bag or something.

later on,
--saul



On Tue, 2008-04-15 at 15:53 -0700, Jody Garnett wrote:
> Here are the kinds of errors reported
> >
> > ArcSDE Error Number        : -21
> > Error Description          : SPECIFIED LAYER IS USE BY ANOTHER USER.
> >
> >  ArcSDE Error Number        : -19
> >  Error Description          : SPECIFIED LAYER ALREADY EXISTS.
> >  Extended Error Description : Error executing stored procedure 
> > sde.layers_util.insert_layer
> >  Extended Error Description : ORA-20019: Layer 829 already exists.
> >
> > WARNING: Out of connections: Timeout waiting for idle object
> > java.util.NoSuchElementException: Timeout waiting for idle object
> >
> My reading of this is that maven testing is much faster than my arcsde 
> instance.
> Jody
> 
> 
> -------------------------------------------------------------------------
> This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
> Don't miss this year's exciting event. There's still time to save $100. 
> Use priority code J8TL2D2. 
> http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
> _______________________________________________
> Geotools-devel mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/geotools-devel



-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
Don't miss this year's exciting event. There's still time to save $100. 
Use priority code J8TL2D2. 
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
Geotools-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/geotools-devel

Reply via email to