From cottrell@wfu.edu Sun Nov  7 10:47:13 2010
From: Allin Cottrell <cottrell@wfu.edu>
To: gretl-devel@gretlml.univpm.it
Subject: Re: [Gretl-devel] nested function error reporting
Date: Sun, 07 Nov 2010 10:47:12 -0500
Message-ID: <Pine.A41.4.58.1011071038530.442864@f1n11.sp2net.wfu.edu>
In-Reply-To: 7337_1288966145_oA5E95Td007580_4CD40FF8.3040709@gmx.net
MIME-Version: 1.0
Content-Type: multipart/mixed; boundary="===============8795789959788346820=="

--===============8795789959788346820==
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 7bit

On Fri, 5 Nov 2010, Sven Schreiber wrote:

> I stumbled across a glitch with error messages occurring in
> nested functions. I'm reporting it not per se, but because it
> might hint at a deeper issue with nested functions, given a
> history of subtle flakiness with function scopes and such -- I
> just don't know [...]

Agreed, the error message(s) you reported are somewhat confusing.
However, I don't think they "hint at a deeper issue" -- it's just
that getting the error reporting "right" is quite tricky. I've now
had another go at this, and I think the output should be clearer.
Here's a simple example:

<script>
function scalar f3 (scalar x)
   y = sqrt(z) # deliberate error here
   return y
end function

function scalar f2 (scalar x)
   return f3(x)
end function

function scalar f1 (scalar x)
   return f2(x)
end function

# main
scalar s = f1(9)
</script>

The output from the above is now:

<output>
? scalar s = f1(9)
The symbol 'z' is undefined

*** error in function f3
> y = sqrt(z)
 called by function f2
 called by function f1
</output>

That is, we print only one "error in function" message, pertaining
to the function which directly produced the error, and if this
function is nested we follow up by printing the "call stack".

Allin

--===============8795789959788346820==--


From svetosch@gmx.net Sun Nov  7 11:21:41 2010
From: Sven Schreiber <svetosch@gmx.net>
To: gretl-devel@gretlml.univpm.it
Subject: Re: [Gretl-devel] nested function error reporting
Date: Sun, 07 Nov 2010 17:21:29 +0100
Message-ID: <4CD6D209.6040601@gmx.net>
In-Reply-To: Pine.A41.4.58.1011071038530.442864@f1n11.sp2net.wfu.edu
MIME-Version: 1.0
Content-Type: multipart/mixed; boundary="===============5260886109610056568=="

--===============5260886109610056568==
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 7bit

Am 07.11.2010 16:47, schrieb Allin Cottrell:


> The output from the above is now:
> 
> <output>
> ? scalar s = f1(9)
> The symbol 'z' is undefined
> 
> *** error in function f3
>> y = sqrt(z)
>  called by function f2
>  called by function f1
> </output>
> 
> That is, we print only one "error in function" message, pertaining
> to the function which directly produced the error, and if this
> function is nested we follow up by printing the "call stack".
> 


looks good! I'll watch out for it showing up in practice.

thanks,
sven

--===============5260886109610056568==--