This problem occurs when there is an mKR syntax error
in the input.  The original case was due to an editing error
on my part -- the token type of "/" was "U" [unknown].

I will investigate further, to determine exactly what happens
during mKE processing of the syntax error.

In the meantime, I will upload the "/" fix to my mirror
web site.  This version includes code to write the results
of each "suspend" level to the mKE .log file.

Dick McCullough
http://mkrmke.org
----- Original Message ----- 
From: "Steve Wampler" <[email protected]>
To: "Richard H. McCullough" <[email protected]>
Sent: Thursday, May 14, 2009 6:45 PM
Subject: Re: [Unicon-group] Windows Unicon 11.3: nested suspend bug ???


> Richard H. McCullough wrote:
>> But parse_file(fd,"nrel","end") returns without
>> executing get_prop() that one last time.  The
>> last execution of get_prop() successfully returns
>>    end relation rname;
>> If I called it once more, it would again succeed
>> by reading & returning the next line following that one,
>> which I don't want it to do.
> 
> Ok - that's fine.  Returning from a procedure before
> exhausting any generators it's using internally is
> just fine.  That's almost certainly not the problem.
> (That behavior is done everywhere in Unicon/Icon programs.)
> 
> There's really too much code in ke to try and track down
> the problem.  (Which is why others are also asking for
> a small code example that illustrates the problem.)
> 
> I really doubt that it's a bug in suspend, so I'd look first
> at the code.
> 
> A couple of less-likely possibilities do exist, however,
> so a few more questions:
> 
>    (1) Are you running the code inside a co-expression
>        [other than the default &main]?
> 
>    (2) Do you end up with deeply nesting recursion during
>        this process?
> 
> If both (1) and (2) are true, you *may* be running afoul
> of the comparatively small stack-allocation size for
> co-expressions.   Overflowing a CE stack can produce
> all sorts of wondrous behavior.  Fortunately it's easy
> to check [well, at least under Linux/Unix, I don't use
> Windows, so you may have to ask Clint what's equivalent
> to bumping up the value of the CEXPRSIZE limit.  [I'm
> getting the name from memory (no book handy), but Clint
> would know the real name from my guess...]
> 
> If you bump it up (way up, to 1MB, say) and the outcome
> is different, then that may be the problem.  I suspect
> you're not running inside a CE, however, so regard this
> as an unlikely cause.
> 
> -- 
> Steve Wampler -- [email protected]
> The gods that smiled on your birth are now laughing out loud.
> 
>

------------------------------------------------------------------------------
Crystal Reports - New Free Runtime and 30 Day Trial
Check out the new simplified licensing option that enables 
unlimited royalty-free distribution of the report engine 
for externally facing server and web deployment. 
http://p.sf.net/sfu/businessobjects
_______________________________________________
Unicon-group mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/unicon-group

Reply via email to