Hi again!
Please excuse if this is a duplicate, my web
mailer choked on that one :/
I've
noticed, that some of the libxml return codes
get
"simplified" when used in xmlreader which can be quite
problematic.
I opened a bug report about this behaviour:
https://bugs.php.net/bug.php?id=55459
The question is, wether
this is considered a bug at all or is
rather seen as a change request.
Throughout the whole xmlreader
extension libxml's 1, 0, -1 return
codes are simply mapped to
true/false.
The fix is trivial, but
I don't know if
changing the return code of these php functions might
break existing
code.
I'm short of a really easy way of working
around
that problem. Of course I could rename the xmlreader extension, fix
it and could load a fixed version, but that seems awkward.
Any
suggestions?
Greetings,
Nico
Hi again!
Please excuse if this is a duplicate, my web
mailer choked on that one :/
I've
noticed, that some of the libxml return codes
get
"simplified" when used in xmlreader which can be quite
problematic.I opened a bug report about this behaviour:
https://bugs.php.net/bug.php?id=55459The question is, wether
this is considered a bug at all or is
rather seen as a change request.Throughout the whole xmlreader
extension libxml's 1, 0, -1 return
codes are simply mapped to
true/false.The fix is trivial, but
I don't know if
changing the return code of these php functions might
break existing
code.I'm short of a really easy way of working
around
that problem. Of course I could rename the xmlreader extension, fix
it and could load a fixed version, but that seems awkward.
Use libxml_get_errors()
-Hannes
Hi!
On Fri, Aug 19, 2011 at 12:27, Nicolai Scheer
scope@planetavent.de
wrote:
I've
noticed, that some of the libxml return codes
get
"simplified" when used in xmlreader which can
be quite
problematic.
Use
libxml_get_errors()
So you're suggesting to call another
function rather than returning something that is already there?
I see how libxml_get_errors() can be used to work around this, but to my
mind, this function should not be called unless I know that in fact there
are errors.
Nevertheless, is there a "real" reason
why this is not implemented? Anyone else feeling uncomfortable with
"calling additional functions to check for error conditions"?
Greetings,
Nico
Hi!
On Fri, Aug 19, 2011 at 12:27, Nicolai Scheer
scope@planetavent.de
wrote:I've
noticed, that some of the libxml return codesget
"simplified" when used in xmlreader which can
be quite
problematic.
Use
libxml_get_errors()So you're suggesting to call another
function rather than returning something that is already there?I see how
libxml_get_errors()can be used to work around this, but to my
mind, this function should not be called unless I know that in fact there
are errors.Nevertheless, is there a "real" reason
why this is not implemented? Anyone else feeling uncomfortable with
"calling additional functions to check for error conditions"?Greetings,
Nico
It is modeled after the .NET XMLReader interface with simple boolean
return values. You can file a feature request to enable exceptions to be
thrown from the reader on invalid XML. i.e. argument to constructor
whether or not to throw exceptions (default false). Right now regardless
of returning false or an int value, you would have to check something to
determine if it were an error or no data (either a function as it is
right now or for a specific return value), so it's always an extra step.
Rob
Hi!
[...]
Nevertheless, is there a
"real" reason
why this is not implemented? Anyone
else feeling uncomfortable with
"calling additional
functions to check for error conditions"?
It is modeled
after the .NET XMLReader interface with simple boolean
return
values. You can file a feature request to enable exceptions to be
thrown from the reader on invalid XML. i.e. argument to
constructor
whether or not to throw exceptions (default false).
Right now regardless
of returning false or an int value, you
would have to check something to
determine if it were an error
or no data (either a function as it is
right now or for a
specific return value), so it's always an extra step.
Of course
it is an extra step both ways. But using libxml_get_errors() forces me to
use libxml_use_internal_errors and libxml_clear_errors afterwards.
Until xmlreader throws exceptions on errors it is not modeled
after the .NET XMLReader, because that part is really missing.
So the
argument chain is somehow broken here :)
Is it possible to
reopen my bug and change it to a feature request or do I have to resubmit
my thoughts?
Greetings,
Nico