Good point. Can you execute the SQL directly against the db?
On Nov 18, 2005, at 11:18 AM, Robert Shubert wrote:
I would definitely try a directdbms in that case, get away from the
SQL generator. The NULL could be an issue. Personally I’m in the
habit of making flags that contain a 0 or a 1 and block the NULL
case. It’s entirely possible that there’s some glitch in dealing
with the NULL column.
On other thing would be to trace the DB. You’d like to know if the
SQL statement is getting to the SQL server or not.
Robert
From: Roland Dumas [mailto:[EMAIL PROTECTED]
Sent: Friday, November 18, 2005 2:04 PM
To: [email protected]
Subject: Re: Witango-Talk: crashing witango 5 on OS X
On Nov 18, 2005, at 9:56 AM, Robert Shubert wrote:
Well, focus on that action then and see if you can determine if the
problem is Witango’s handling of the action, or the database or the
connectivity driver.
I have been through the ODBC dramas and know their signatures -
which show up in the crash log. This has a different signature, and
looks like it's wholly within witango, as all the threads say
"witango" on them.
Again, I would start by either moving that search action to a TCF
call to further isolate it or use a Direct DBMS to remove Witango’s
SQL generator from the picture.
You may want to start logging the input that goes into your search
criteria as before write a file, send an email, whatever, just
record all the criteria that went into the statement. After you get
a good response you can mark the criteria set as OK or delete it.
When the crash happens, you’ll theoretically be left with the bad
criteria to do further testing with.
search criteria are UID = <@ARG UID>. It's looking up a transaction
record for processing.
thing is that this worked without a hitch until I brought in one
more field. The field contains a null or a Y. That's it.
Robert
From: Roland Dumas [mailto:[EMAIL PROTECTED]
Sent: Friday, November 18, 2005 12:23 PM
To: [email protected]
Subject: Re: Witango-Talk: crashing witango 5 on OS X
On Nov 16, 2005, at 1:23 PM, Robert Shubert wrote:
Roland,
I’ve seen crashing issues move from 5.0 ro 5.5 on Windows, so there
really are no guarantees. That doesn’t change the fact that 5.5
(especially on 10.4) seems to be a much better solution.
that said…
The first thing you need to do is isolate where in the TAF it’s
crashing (if it’s always the same). You’re best bet is to add
either a file write action, email action, or a database insert
action between each major point of the TAF. You can start with just
a few, maybe even just one in the middle. But eventually you should
(hopefully) be able to isolate it down to a particular action.
The _function has six actions. I've isolated to the single action
that is the culprit. A relatively simple search action with a left
outer join. Microscope to the XML reveals nothing out of the
ordinary. No funny business at all.
I would first try deleting that action and recreating from scratch.
done. Visually, the XML from the 'old' and the 'new' action are
identical.
There have been some discussions about TAF corruption. Try
searching for !CST, and look back through the WT list for more info.
I've had those experiences, but nothing like that here.
If the TAF in question isn’t too complex, you might just want to
recreate it. Just don’t copy any code over, just to be safe.
it would be a chunk of work, and since the crashes are associated
with only one action in one function, I'd like to get to the cause
within that small area, rather than recreate it from scratch and
rebuild the same crash.
Make sure you send all the crash info to WT. You can ask them if
there’s any hints to the problem. If the crash info isn’t
corrupted, they can usually see if it’s a TAF, TCF, datasource,
cache, etc. issue - although I make no guarantees here.
their suggestion was look at the log. Unfortunately the witango log
doesn't capture the crash, nor anything near it. Big black holes in
the log.
Robert
From: Roland Dumas [mailto:[EMAIL PROTECTED]
Sent: Wednesday, November 16, 2005 3:59 PM
To: WiTango List List
Subject: Witango-Talk: crashing witango 5 on OS X
Have had this particular crash mode for a few months and not been
able to figure it out at all. It kills the witango log, so there is
nothing there. The crash happens when one application is hit, and
occurs randomly. it might occur 1 in 100 hits. What the taf is
doing is just a few queries, nothing fancy.
the thread that crashes looks like this:
Thread 23 Crashed:
0 witangod 0x00068ef8 0x1000 + 0x67ef8
1 witangod 0x000482a8 0x1000 + 0x472a8
2 witangod 0x00069cc4 0x1000 + 0x68cc4
3 witangod 0x0006b300 0x1000 + 0x6a300
4 witangod 0x00019bdc 0x1000 + 0x18bdc
5 witangod 0x0001a1d4 0x1000 + 0x191d4
6 witangod 0x00017e0c 0x1000 + 0x16e0c
7 witangod 0x0001ac90 0x1000 + 0x19c90
8 witangod 0x0014e7d8 0x1000 + 0x14d7d8
9 libSystem.B.dylib 0x90024910 _pthread_body + 0x28
any clues where to look for this.
(pls don't say upgrade to 5.5. The client has not had enough
trouble free service from witango to be confident in another version.)
______________________________________________________________________
__TO UNSUBSCRIBE: Go to http://www.witango.com/developer/
maillist.taf__________________________________________________________
______________TO UNSUBSCRIBE: Go to http://www.witango.com/
developer/maillist.taf
Roland A. Dumas
310 W. Bellevue Ave.
San Mateo, CA 94402
650-347-1373
415-412-9300 (cell)
AIM: radumas
[EMAIL PROTECTED]
______________________________________________________________________
__ TO UNSUBSCRIBE: Go to http://www.witango.com/developer/
maillist.taf
______________________________________________________________________
__ TO UNSUBSCRIBE: Go to http://www.witango.com/developer/maillist.taf
Bill
William M. Conlon, P.E., Ph.D.
To the Point
345 California Avenue Suite 2
Palo Alto, CA 94306
vox: 650.327.2175 (direct)
fax: 650.329.8335
mobile: 650.906.9929
e-mail: mailto:[EMAIL PROTECTED]
web: http://www.tothept.com
________________________________________________________________________
TO UNSUBSCRIBE: Go to http://www.witango.com/developer/maillist.taf
