Yes, missing typemap entries will trigger uninit warnings in Inline::CPP; I
have seen it many times. I ended up patching my copy to report what type
conversion was failing. I may have even filed a bug; I can't remember. It's
a little tangled up in a problem with error reporting being mishandled.
But why are you using double* instead of a plain double? double* is
ambiguous; you'll need to explain to it somehow what you mean by that. A
return value? An array? A pointer to a single double to make it fit nicely
into a single word? An image buffer that is really an array of RGB triples
in which you are supposed to draw the answer in a comical font?
I'm surprised at dgamlm(xmin, xmax). I didn't realize XS knew about
references.
On Tue, Jun 9, 2009 at 7:11 AM, ARTHUR GOLDBERG <a...@cs.nyu.edu> wrote:
> Thanks much Rob
>
> The versions are
> $ perl -le 'require Inline;print $Inline::VERSION'
> 0.45
> $ perl -le 'require Parse::RecDescent;print $Parse::RecDescent::VERSION'
> 1.94
>
> FWIW my sense is that the uninitialized hash value leads to a missing entry
> in the typemap for 'double *'. I can recreate the compilation error with
> this test code, which works fine with the C++ function addDouble removed:
>
> #!/usr/bin/perl
>> use warnings;
>> use lib '/opt/local/lib/perl5/site_perl/5.8.9/';
>>
>> print "9 + 16 = ", add(9, 16), "\n";
>>
>> use Inline CPP => <<'END';
>>
>> int add(int x, int y) {
>> return x + y;
>> }
>>
>> double addDouble( double *x, double *y ) {
>> return x[0] + y[0];
>> }
>>
>> END
>>
>>
>
> Below are lines 875 ff from
> /private/var/vpfiles/_Inline/build/VirtualPlant/Analysis/Correlation/CorrelationCalc_fa1b/CorrelationCalc_fa1b.xs
>
>> double
>> BetaI(x, pin, qin)
>> double x
>> double pin
>> double qin
>>
>> double
>> LnBeta(a, b)
>> double a
>> double b
>>
>> double
>> d9lgmc(x)
>> double x
>>
>> double
>> LogRel(x)
>> double x
>>
>> double
>> LnGamma(x)
>> double x
>>
>> double
>> Gamma(x)
>> double x
>>
>> int
>> initds(os, nos, eta)
>> double * os
>> int nos
>> double eta
>>
>> double *
>> getFlArray(sv)
>> SV * sv
>>
>> void
>> dgamlm(xmin, xmax)
>> double & xmin
>> double & xmax
>> PREINIT:
>> I32 * __temp_markstack_ptr;
>> PPCODE:
>> __temp_markstack_ptr = PL_markstack_ptr++;
>> dgamlm(xmin,xmax);
>> if (PL_markstack_ptr != __temp_markstack_ptr) {
>> /* truly void, because dXSARGS not invoked */
>> PL_markstack_ptr = __temp_markstack_ptr;
>> XSRETURN_EMPTY; /* return empty stack */
>> }
>> /* must have used dXSARGS; list context implied */
>> return; /* assume stack size is correct */
>>
>> double
>> dcsevl(x, cs, n)
>> double x
>> double * cs
>> int n
>>
>> double
>> Dint(d)
>> double d
>>
>>
>>
>
> Also, ultimately I think it would be great to modify the Inline CPP's
> Makefile.PL so that it selects the right compiler and library on Mac OS X.
>
> BR
> A
>
>
> On Jun 9, 2009, at 9:07 AM, Sisyphus wrote:
>
>
>> ----- Original Message ----- From: "ARTHUR GOLDBERG" <a...@cs.nyu.edu>
>>
>> and the compilation finally dies with
>>>
>>>> /usr/bin/perl /System/Library/Perl/5.8.8/ExtUtils/xsubpp -typemap /
>>>> System/Library/Perl/5.8.8/ExtUtils/typemap CorrelationCalc_fa1b.xs
>>>> > CorrelationCalc_fa1b.xsc && mv CorrelationCalc_fa1b.xsc
>>>> CorrelationCalc_fa1b.c\nError: 'double *' not in typemap in
>>>> CorrelationCalc_fa1b.xs, line 904\nError: 'double *' not in typemap
>>>> in CorrelationCalc_fa1b.xs, line 910\nError: 'double *' not in
>>>> typemap in CorrelationCalc_fa1b.xs, line 932\nmake: ***
>>>>
>>>
>> I should know how to generate that error ... but apparently I don't.
>> Could you provide us with lines 904, 910 and 932 of
>>
>> /private/var/vpfiles/_Inline/build/VirtualPlant/Analysis/Correlation/CorrelationCalc_fa1b/CorrelationCalc_fa1b.xs
>> (and maybe a few lines either side for context, if you think that might
>> help).
>>
>> Also, what version of Inline, and what version of Parse::RecDescent do you
>> have:
>> perl -le 'require Inline;print $Inline::VERSION'
>> perl -le 'require Parse::RecDescent;print $Parse::RecDescent::VERSION'
>>
>> Cheers,
>> Rob
>>
>>
>>
>>
>
