Re: Forms 3 Problem - Please Help

[Quoted] If you don't have write permission to the current directory, no core file will be produced. I guess that's why there isn't a core file - I assume that your user exit doesn't call exit() :-)

If you have the source for the user exit, then you should be able to make better use of the core file once you get it: 1) Compile your runform with the option to include debugging information (as far as I remember that the -g option, but better check yourself) 2) copy the newly generated runform to the server where the problem appears 3) run it and produce a core file. If it does, expect a delay of a couple of seconds as core files from SQL*Forms can be rather sizeable. 4) examine the core file - hopefully you will be able to produce a stack trace, which should help you pinpoint the problem.

What platform are you using? Different platforms have different ways of examining core files - look for commands like xdb, dbx etc...

Hope this helps

--
Karl E. Jorgensen
Return address munged to prevent spam, use k a r l _at_ j o r g e n s e n . c o
m instead...

P Seely wrote in message <378b41be.7338686_at_allnews.nbnet.nb.ca>...


>On Tue, 13 Jul 1999 08:30:25 +0200, "Johan Wegener" <xjw_at_xdde.xdk>


>wrote:

>

>><pseely_at_nbnet.nb.ca> skrev i en

>>nyhedsmeddelelse:378a9da1.82371303_at_allnews.nbnet.nb.ca...

>>[snip]

>>> The user_exit calls to the library functions work fine on the

>>> Developent server.  When the a code release to the Production server

>>> was attempted, the thing does.  When you get to the point where the

>>> user_exit call is made, you can see the result flash briefly on the

>>> screen in place (telling me the call succeeds) but then the form craps

>>> out - your prompt appears.  Forms is not dumping core (which would be

>>> nice, IMHO, since there'd be something to work with.)

>>[snip]

>>

>>Peter,

>>alas, I am not sure exactly what you mean from your description, could you

>>be a bit more specific as to how this error shows itself?

>>

>>Furthermore could you enlighten me to some details:

>>1) How do you exchange values to and from the user exit to the Form?

>>2) Do you use the NO_SCREEN parameter when you call your user exit?

>>2) What kind of terminal or terminal emulation are you attempting, is the

>>emulation/terminal the very same on both systems?

>>3) What kind of resource (Oracle*Terminal) file are you using with Forms?

Is


>>it identical on both systems?


>>

>>Regards,

>>Johan

>>

>>

>

>Well, the error doesn't show itself.  What happens is that when you

>tab to the field, the user exit is called.  The user exit actually

>posts the result to the database, after which the form retrieves the

>results from the database into form variables.  You see the result

>appear in the field where it should be for about a half a second, then

>your command prompt appears and the form is dead.  There is no error

>file, no core dump, no nothing.

>

>1) The user exit posts the results of its processing to database

>tables from which the form retrieves and displays them.

>

>2) No, we're not using NO_SCREEN when calling user_exit.  We use it

>when we do a HOST command though.  In this case (calling the third

>party library) we're doing user_exit.

>

>3) We're doing vt220 terminal emulation.  It is the same on both

>systems.

>

>4) We have a resource file which is used for Oracle*Terminal.  It is

>the same on both systems.

>

>Interestingly enough, I found out that our system architect tried to

>use a msg_popup in this procedure to pinpoint where the process was

>at, and that caused the form to crap out in the exact same manner.  It

>seems that attempting to do any output to the screen at this point

>causes the form to die.

>

>We are able to query the database, though.  This doesn't seem to be a

>problem.  I also feel compelled to say that,  in another environment

>on this Production server (which is running an older version of this

>third-party library) this form and all other calls to the library work

>fine.

>

>I hope this information is useful and appreciate any light you can

>shed on this.

>

>Thanks a lot

>Peter

>

>

>