Re: spurious 'if' error after fault

[sfeam (E. Merritt) <eam...@gm...>]

On Thursday, 13 October 2011, sfeam (Ethan Merritt) wrote:
> On Thursday, 13 October 2011, pl...@pi... wrote:
> > Hi,
> > 
> > I don't know exactly what the function eval error is but the break in 
> > execution seems to highlight a bug in the new if/then/else error trapping.
> > 
> > Seems the parser is set half way though an if block and then gets 
> > confused when I reload the file.

Never mind. I think I see the issue.
The "Undefined value during function evaluation" message comes from
"fit". The "fit" subsystem has its own error-handling routine error_ex()
rather than calling int_error() like the rest of the program.
But error_ex() does not do all the cleanup that int_error() does.
We need to either add calls to the various cleanup routines in error_ex()
or modify "fit" to call int_error() like everyone else. 

	Ethan



> > 
> > Otherwise the new syntax makes life a lot easier. Thanks.
> > 
> > Don't know if this is important .
> > 
> > regards. Peter.
> > 
> > 
> > 
> > gnuplot> load "doublet_integral.gnu
> >           Undefined value during function evaluation
> 
> disregarding this one for a moment (although of course it might be crucial)...
>  
> > gnuplot> load "doublet_integral.gnu
> > 
> > gnuplot> if (use_SD) outfilebase=outfilebase."SD_fit."
> >                       ^
> >           "doublet_integral.gnu", line 65: Old-style if/else statement 
> > encountered inside brackets
> > gnuplot> load "doublet_integral.gnu
> >           Undefined value during function evaluation
> 
> I don't know how to make that first error message any clearer, but I'm
> open to suggestions.   An old-style "if" with no { ... } clause
> causes the entire rest of the current bracketed clause to be skipped.
> Given that an unknown number of input commands were skipped, 
> it's hardly surprising that error messages about things like
> undefined values might occur later.
> 
> There is currently a bug (at least I consider it a bug) in the
> "load" command that might also be relevant.  If you nest several
> load commands, and the deepest one says "exit", it is supposed to 
> back out of this level and resume execution at the previous level.
> Instead the current cvs version falls back all the way to the top
> level of execution.  I don't know exactly when this bug crept in,
> but it's been there for a while now.
> 
> If you think your scripts demonstrate a failure not explained by
> either of these factors, please attach them to a bug report on
> SourceForge so that we can have a look at it.
> 
> 
> ------------------------------------------------------------------------------
> All the data continuously generated in your IT infrastructure contains a
> definitive record of customers, application performance, security
> threats, fraudulent activity and more. Splunk takes this data and makes
> sense of it. Business sense. IT sense. Common sense.
> http://p.sf.net/sfu/splunk-d2d-oct
> _______________________________________________
> gnuplot-beta mailing list
> gnu...@li...
> https://lists.sourceforge.net/lists/listinfo/gnuplot-beta
>