From ramises.silva at kryptus.com  Tue Oct 17 11:41:34 2017
From: ramises.silva at kryptus.com (=?UTF-8?Q?Ramis=C3=A9s_Martins_da_Silva?=)
Date: Tue Oct 17 13:49:11 2017
Subject: [studxml-users] Asserts in next_expect functions
Message-ID: <CAH8pVaOq+4WqAVBrjpiEp2j+jCQ+fXLcUX29+uUNArg7Nrn8cg@mail.gmail.com>

Hi,

I'm curious about the "next_expect functions that also set the content". In
it's implementation it is asserted that the event must be a start_element.
My question is if the placement of this assert inside the function is
important. For me it seems more natural that a blocking condition like an
invalid parameter test comes first in the function.

-- 
*Ramis?s Martins*
From boris at codesynthesis.com  Wed Oct 18 10:59:10 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Oct 18 10:59:19 2017
Subject: [studxml-users] Asserts in next_expect functions
In-Reply-To: <CAH8pVaOq+4WqAVBrjpiEp2j+jCQ+fXLcUX29+uUNArg7Nrn8cg@mail.gmail.com>
References: <CAH8pVaOq+4WqAVBrjpiEp2j+jCQ+fXLcUX29+uUNArg7Nrn8cg@mail.gmail.com>
Message-ID: <boris.20171018165802@codesynthesis.com>

Ramis?s Martins da Silva <ramises.silva@kryptus.com> writes:

> I'm curious about the "next_expect functions that also set the content". In
> it's implementation it is asserted that the event must be a start_element.
> My question is if the placement of this assert inside the function is
> important. For me it seems more natural that a blocking condition like an
> invalid parameter test comes first in the function.

Yes, I agree, and am not sure why we did it this way. Fixed for the
next release, thanks!

Boris

From pedro at kryptus.com  Wed Oct 18 13:52:08 2017
From: pedro at kryptus.com (Pedro Calixto)
Date: Wed Oct 18 13:52:22 2017
Subject: [studxml-users] Asserts in next_expect functions
In-Reply-To: <boris.20171018165802@codesynthesis.com>
References: <CAH8pVaOq+4WqAVBrjpiEp2j+jCQ+fXLcUX29+uUNArg7Nrn8cg@mail.gmail.com>
	<boris.20171018165802@codesynthesis.com>
Message-ID: <93087730-2225-d1cf-7f58-689feeefdf94@kryptus.com>

There is a difference, however, between these assertion placements.

To test it, I moved the assertion found in /inline void
parser::next_expect (event_type e, const std::string& ns, const
std::string& n,?//content_type c)/ (parser.ixx) to the beginning of this
function.


The parsed xml was:

<?xml version="1.0" encoding="utf-8"?>
<person id="123">
? <name>John Doe</name>
</person>


The code that tested it was:

int main(int argc, char *argv[]) {

? std::ifstream ifs(argv[1]);
? xml::parser p(ifs, argv[1]);

? std::string const Name("animal");
? std::string const Ns("");
? auto Content = xml::content(xml::content::complex);
? p.next_expect(xml::parser::end_element, Ns, Name, Content);

? return 0;
}


Before moving the assertion to the beginning, there was an exception
thrown (parsing), and the program output was:

terminate called after throwing an instance of 'xml::parsing'
? what():? error.xml:2:0: error: end element 'animal' expected
Aborted (core dumped)

After moving the assertion to the beginning, there was no exception, but
the assertion error:

main: /usr/local/include/xml/parser.ixx:136: void
xml::parser::next_expect(xml::parser::event_type, const string&, const
string&, xml::parser::content_type): Assertion `e == start_element' failed.
Aborted (core dumped)

I wonder which alternative would be better for displaying the error to
the user, since the exception description feels more clear about what
was wrong (IMO).

-- 
Pedro Calixto,
Estagi?rio
KRYPTUS EED S/A
Trust in Cybersecurity
+55 19 3112 5000
www.kryptus.com

From boris at codesynthesis.com  Thu Oct 19 13:40:28 2017
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Thu Oct 19 13:40:36 2017
Subject: [studxml-users] Asserts in next_expect functions
In-Reply-To: <93087730-2225-d1cf-7f58-689feeefdf94@kryptus.com>
References: <CAH8pVaOq+4WqAVBrjpiEp2j+jCQ+fXLcUX29+uUNArg7Nrn8cg@mail.gmail.com>
	<boris.20171018165802@codesynthesis.com>
	<93087730-2225-d1cf-7f58-689feeefdf94@kryptus.com>
Message-ID: <boris.20171019193705@codesynthesis.com>

Pedro Calixto <pedro@kryptus.com> writes:

> There is a difference, however, between these assertion placements.

The assertion checks the function pre-condition which is that it should
only be called for start_element events. To put it another way, if the
assertion is triggered, then the code that called next_expect() is
invalid. If the exception is thrown, then the XML provided by the
user is invalid. In most cases you want to find out that your code
is broken as soon as possible.

Boris