[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-kdoc
Subject:    Re: Small bugs
From:       Hans Martens <HMartens () jasongeo ! com>
Date:       2000-05-02 13:11:02
[Download RAW message or body]

> > 1) It fails to parse inline template constructors. E.g.
> > 
> > <snip>
> 
> Can you try whether this patch fixes it?
> 
> --- kdoc.3      Mon Feb 14 12:31:25 2000
> +++ kdoc        Tue May  2 13:59:45 2000
> @@ -659,7 +659,7 @@
> my $pure = $4 eq "" ? 0 : 1;
> 
> if ( $tpn =~ /((?:\w+\s*::\s*)?operator.*?)\s*$/        #operator
> -                                       || $tpn =~  \
> /((?:\w*\s*::\s*~?)?[-\w:]+)\s*$/ ) { # normal +                               || \
> $tpn =~ /((?:\w*(?:<[\w\s_,]*>)?\s*::\s*~?)?[-\w:]+)\s*$/ ) { # normal my $name = \
> $1; $tpn = $`;
> $newNode = newMethod( $tpn, $name,

Tried that. Works partially. For one class i.s.o. "'KeyedDataSels' has
no KidAccess", I now get

findRef: no name at
/home/hans/tools/Any/kdoc/kdoc-2.0a35/share/kdoc/kdocAstUtil.pm line 101
        kdocAstUtil::findRef('Ast=HASH(0x222ad4)', '') called at
/home/hans/tools/SunOS/bin/kdoc line 1177
        main::newMethod('template < class KeyType, class DataType >
KeyedDataSels ', '< KeyType, DataType > :: KeyedDataSels', '', 0, 0)
called at /home/hans/tools/SunOS/bin/kdoc line 664
        main::identifyDecl('template < class KeyType, class DataType >
KeyedDataSels < KeyTy...') called at /home/hans/tools/SunOS/bin/kdoc
line 259
        main::parseFiles() called at /home/hans/tools/SunOS/bin/kdoc
line 161
*** Error code 2

Code for this class is e.g.

template < class KeyType, class DataType >
KeyedDataSels < KeyType, DataType > :: KeyedDataSels()
{}

template < class KeyType, class DataType >
KeyedDataSels < KeyType, DataType > :: KeyedDataSels(const KeyedDataSels
<KeyType, DataType> &kds)
{
	copy(kds);
}

> Gee! I have approx 10 years of C++ experience, but I didn't know such a
> thing was possible. I'd say, that's perverse, so I will certainly not work
> on that ;-) You may try yourself, but a look at the respective function in
> kdoc reveals that this is probably not so simple.
> 
> Unrelated question: why do you write such a line in the beginning?
I agree with you completely. However, there is such a thing as legacy
code... (Some programmers seem to feel that the quality of a program is
inversely proportional to its ASCII code length ;-)

Cheers,
Hans

BTW, I fixed some other small problems which you may or may not have
found:

1) file 'kdoc', line 108, parsing the cppcmp option should read

	"cppcmd|C=s",	\$cppcmd,

2) Since I sometimes get empty HREF's in my HTML documentation, file
'kdocHTMLutil.pm', subroutine 'sub-hyper' changed to:

sub hyper
{
    confess "hyper: undefed parameter $_[0], $_[1]"
	unless defined $_[0] && defined $_[1];
    my $out = "";
    if ($_[0] ne "") { ## filter out empty refs
    	$out = "<A HREF=\"$_[0]\">".esc($_[1])."</A>";
    }
    else {
    	$out = esc($_[1]);
    }
    return $out;	
}

3) For similar reasons, in subroutine 'writeNode' in file 'kdocLib.pm',
change

		print LIB "<C NAME=\"", $n->{astNodeName},
			"\" REF=\"$prefix.html\">\n";

to

    	if( $prefix ne "" ) { ## filter out empty refs
    		print LIB "<C NAME=\"", $n->{astNodeName},
	    		"\" REF=\"$prefix.html\">\n";
        }


4) Similarly, further down in the same subroutine:

            print LIB "<$type NAME=\"", $kid->{astNodeName},
                "\" REF=\"$prefix.html#", $kid->{astNodeName}, "\">\n";

to

    if ($prefix ne "") { ## filter out empty refs
            print LIB "<$type NAME=\"", $kid->{astNodeName},
                "\" REF=\"$prefix.html#", $kid->{astNodeName}, "\">\n";
    }

I have not really checked what causes the empty refs, I'm afraid, I
removed them "phenomenologically".


[prev in list] [next in list] [prev in thread] [next in thread]