Recent changes to support-requests

Fix for locale-dependent floating point parsing

Harri Porten — Tue, 15 Aug 2017 19:33:25 -0000

atof() works depending on the locale. The usage of periods is hardcoded, though.

Proposed patch attached.

Bug in test.xml

Alexandre Moine — Thu, 30 Jul 2015 20:00:13 -0000

Hi,

I tried to make a package for Fedora, and the binary return some errors (see out.htm), I investigated the error, and the problem came from an inversion of values in comment of test.xml. I corrected it, and all work fine.

I added a patch.

Thank you

Bug in lex parser

Anonymous — Wed, 27 Mar 2013 20:59:24 -0000

There's a bug in the lex parser currently with validating the XPath expression. If your XPath element name contains any of the key words used by the lexical expressions, they get flagged as invalid XPath. For example:

XPath:

//star

XML:

<galaxy>
<star id="A001">This is a valid element</star>
</galaxy>

I removed div from the set for my specific application, but any keyword (mod, add, div, star), etc. will be flagged as an invalid XPath syntax.

Assignment operator should be return by reference

Gaurav — Wed, 19 Dec 2012 05:59:11 -0000

in File tinyxml.h
In TiXmlHandle class assignment operator should return by reference.

Here is diff.

- TiXmlHandle operator=( const TiXmlHandle& ref ) { this->node = ref.node; return *this; }
+ TiXmlHandle& operator=( const TiXmlHandle& ref ) { this->node = ref.node; return *this; }

Memory Leak in TinyXPath

Benjamin Rood — Mon, 04 Jun 2012 18:40:54 -0000

If a node set is computed more than once for an XPath expression, a memory leak occurs because the "operator=" overloaded function in node_set.cpp does not check for a previous allocation of the "vpp_node_set" and "op_attrib" members of the object and free them appropriately.

I managed to reproduce this consistently via the following test:

1.) Load a document
2.) Compute a node set via the "u_compute_xpath_node_set()" function repeatedly, such as in the condition of a for loop

Example code:

[code]

TiXmlDocument* document = new TiXmlDocument( "test.xml" );
document->LoadFile();

TinyXPath::xpath_processor proc( document->RootElement(), "/Test/*/text()" );

for ( unsigned int i = 0; i < proc.u_compute_xpath_node_set(); i++ )
{
TiXmlNode* node = proc.XNp_get_xpath_node( i );
printf( "%s\n", node->value() );
}

[code]

I have attached a patch that solves this memory leak.

Win64 and TinyXml 2.6.2

Malcolm Nixon — Fri, 25 Nov 2011 06:34:42 -0000

Currently TinyXPath emits compile warnings for Win64 targets (size_t to unsigned conversions) and contains a fairly elderly version of TinyXml. Attached is a patch file that adds appropriate casts and upgrades TinyXml to the latest version.

last() -1 returns last element

sulc — Fri, 05 Aug 2011 07:57:49 -0000

calling /root/node[last()]/@id does not work as expected. /root/node[last() - 1]/@id returns last element.

xml:
<root>
<node @id = '0001'>
...
</node>
...
<node @id = '0005'>
...
</node>
</root>

text() content comparism doesn't work

Anonymous — Fri, 13 Jun 2008 11:44:33 -0000

hi,

i tried to compile some query like
/people/person[name/text()='Dude']
(compare http://www.zrinity.com/xml/xpath/xpath.cfm\)
with TinyXPath::er_compute_xpath() and I got an error number 3 (execution error) instead of my result.

as long as I don't use text='xyz' it works..

please help and/or patch ;)