erroneous error codes

[Richard Gaskin]

Wilhelm Sanke wrote:
> Yesterday I had written:
> 
>> Errordialog is indeed defined in the engine (make a textsearch of the 
>> engine with a word processor), which means that contrary to what 
>> Jacqueline quoted from Tuviah (see above) there must have been changes 
>> with respect to error codes in the engine.

Yes, errorDialog is the message sent by the engine to notify the 
environment of script execution errors.

The handler for this message is in MC's backscript (btn "newObj" of 
stack "MetaCard Menu Bar"):

on errorDialog which
   set the executionerror of card 1 of stack "Execution Error" to which
   modeless "Execution Error"
   send "refresh" to card 1 of stack "Execution Error"
end errorDialog

The error messages are stored in a hidden field in the "Execution Error" 
  window.  These messages had changed once before, and after complaining 
to Scott they have never changed since.  Tuv maintains that these have 
not changed in Rev v2.2.


> After a second look with some more time at hand I am less sure about the 
> exactness of my diagnosis.
> 
> First of all, there are two (identical?) script editors in our IDE both 
> of which also use "errordialog" much in the same way as in stack 
> "mctools" but additionally check for the script limits of the Starter 
> Kit. One of these stacks, "script editor", is by default invisible, but 
> both stacks "script editor" and "script editor 1" seem to contain the 
> same scripts; however, I have not done a word-for-word comparison.

The MC IDE clones the "script editor" stack when editing a script.  I 
had in advertently saved the stack when I posted the last build.

The presence of additional script editor clones apparently has no affect 
on behavior; it's simply one less step the IDE performs when editing a 
script.  For cleanliness, however, I agree that it should be removed 
from the final build.

Going forward, with the approval of the folks here I'll add a 
"StripAndShip" handler in the IDE to run before posting which will take 
care of such anomalies. Agreed?


> As a substack of the home stack we have got a "Stack 988758648" which 
> seems to be entirely empty. Where does it come from?

I believe it was in an old Home stack downloaded from metacard.com, from 
which the v2.5.1 and v2.6 releases were made.  It seems to serve no 
purpose and can be deleted in the next build.


> ====================
> 
> I re-checked - I think others in our group have done that already - the 
> identity of the MC_IDE stacks on my Windows computer from 2.5.1, 2.6a2 
> to the different versions of 2.6. With the exception of the different 
> engines the IDE stacks are absolutely indentical. But indeed for 
> versions higher than 2.5.1 the said *first* error dialog with "integer" 
> occurs.
> However this does not hold for standalones, which use the same engines.

Good check -- thanks for that.

> I then checked the latest engine Rev 2.2 of April 4 in both IDEs; we get 
 > the "integer" error in the MC_IDE, but not in Revolution.

Verified.  This may be the result of changes to the Rev error dialog to 
accomodate changes in the engine.  I have requested a summary of engine 
error-handling changes from Tuv; it's not yet arrived, but he may just 
be busy.

> I hope I am not too far off the mark with a "logical" conclusion that if 
> the error codes of the engines have not been changed then something else 
> must have been changed that pertains to the interaction of the engine 
> with the MC_IDE - or does anybody see other possible conclusions on the 
> basis of the described facts?

That would be correct if the premise that error codes have not changed 
in the engine is correct.  While that has been reported to be the case 
by Tuv, it may not be entirely accurate.

There are only two factors at play, the engine and the IDE.  The IDE 
performs as expected with engine v2.5.1, but the same IDE gives 
incorrect error messages with engine v2.6.  Jacque has verified this as 
well.

So either our test with the MC IDE 2.6b1 and engine v2.5.1 was somehow 
done erroneously, or this would seem to indicate that a change occurred 
in the engine.

Do you have a recipe for generating a false error message reliably?  I 
spent a little time on it over the weekend while preparing the beta 
release but was unable to get a consistent result.

-- 
  Richard Gaskin
  Fourth World Media Corporation
  ___________________________________________________________
  Ambassador at FourthWorld.com       http://www.FourthWorld.com