Re: svn commit: r676161 - /xmlgraphics/fop/trunk/src/java/org/apache/fop/events/model/EventModel.java

2008-07-13 Thread Andreas Delmelle

On Jul 13, 2008, at 04:07, Peter B. West wrote:


Andreas Delmelle wrote:


URL and URI are both used here. But I think URL is the  
mandatory term

here. The other thing is URI Syntax which does not refer to URI
itself. Since a URL is a URI, but not all URIs are URLs, I  
believe your

example above is slightly incorrect.
Nice catch. For the moment, better to be safe than sorry. Indeed,  
a URI could also be something not referring to a file (or, better  
put: a valid input source for a transformation)
Strictly speaking, according to the above definition, the systemId  
is allowed to be a 'mailto:' (?) :-/

Sloppy editors... ;-)


I don't think so. The docs for java.net.URL include
The syntax of URL is defined by RFC 2396: Uniform Resource  
Identifiers (URI): Generic Syntax, amended by RFC 2732: Format for  
Literal IPv6 Addresses in URLs.


That is, the syntax for URLs is defined as part of the URI syntax.


Yes, that's the point, but 'URI Syntax' is more general. All URLs are  
URIs, but URIs can also be opaque, like a mailto:, news:, isbn:...


The point is that the API docs of  
javax.xml.transform.stream.StreamSource do not explicitly limit the  
parameter to URLs. They do indicate that the constructor is meant to  
create a new StreamSource off a URL, but it does not even hint at the  
(common-sense) requirement that the parameter String actually points  
to a valid, live resource that can be an input source for a  
transformation.


No immediate problem for us, but it just seems a bit sloppy on the  
side of the maintainers of those docs...



Andreas


Re: svn commit: r676161 - /xmlgraphics/fop/trunk/src/java/org/apache/fop/events/model/EventModel.java

2008-07-12 Thread Andreas Delmelle

On Jul 12, 2008, at 14:19, [EMAIL PROTECTED] wrote:


Author: jeremias
Date: Sat Jul 12 05:19:40 2008
New Revision: 676161

URL: http://svn.apache.org/viewvc?rev=676161view=rev
Log:
Attempt to fix a potential build problem.


FWIW, I've locally replaced all occurrences of File.toURL() in the  
codebase to File.toURI().toURL(). Once I've confirmed this breaks no  
tests, I'll commit the changes, so this is out of the way.


Going through the occurrences, I'm getting the impression that in  
some cases, we don't even really need the URL. The URI would do just  
fine if we don't need the functionality for opening a stream... Maybe  
in this particular case, passing through a URL could also be avoided  
(?) If the error is generated by the used StreamResult implementation  
calling File.toURL(), then something like:


Result res = new StreamResult(outputFile.toURI().toASCIIString());

would already be enough (?)


Andreas


Re: svn commit: r676161 - /xmlgraphics/fop/trunk/src/java/org/apache/fop/events/model/EventModel.java

2008-07-12 Thread Jeremias Maerki
On 12.07.2008 14:49:03 Andreas Delmelle wrote:
 On Jul 12, 2008, at 14:19, [EMAIL PROTECTED] wrote:
 
  Author: jeremias
  Date: Sat Jul 12 05:19:40 2008
  New Revision: 676161
 
  URL: http://svn.apache.org/viewvc?rev=676161view=rev
  Log:
  Attempt to fix a potential build problem.
 
 FWIW, I've locally replaced all occurrences of File.toURL() in the  
 codebase to File.toURI().toURL(). Once I've confirmed this breaks no  
 tests, I'll commit the changes, so this is out of the way.

Thanks a lot.

 Going through the occurrences, I'm getting the impression that in  
 some cases, we don't even really need the URL. The URI would do just  
 fine if we don't need the functionality for opening a stream... Maybe  
 in this particular case, passing through a URL could also be avoided  
 (?) If the error is generated by the used StreamResult implementation  
 calling File.toURL(), then something like:
 
 Result res = new StreamResult(outputFile.toURI().toASCIIString());
 
 would already be enough (?)

In most cases, yes. But there's really a difference between a URI and a
URL. But what's confusing is the following (the Javadocs for
StreamSource):

quote
public StreamSource(String systemId)
Construct a StreamSource from a URL.
Parameters:
systemId - Must be a String that conforms to the URI syntax.
/quote

URL and URI are both used here. But I think URL is the mandatory term
here. The other thing is URI Syntax which does not refer to URI
itself. Since a URL is a URI, but not all URIs are URLs, I believe your
example above is slightly incorrect.

Jeremias Maerki



Re: svn commit: r676161 - /xmlgraphics/fop/trunk/src/java/org/apache/fop/events/model/EventModel.java

2008-07-12 Thread Peter B. West

Jeremias Maerki wrote:

On 12.07.2008 14:49:03 Andreas Delmelle wrote:

On Jul 12, 2008, at 14:19, [EMAIL PROTECTED] wrote:


Author: jeremias
Date: Sat Jul 12 05:19:40 2008
New Revision: 676161

URL: http://svn.apache.org/viewvc?rev=676161view=rev
Log:
Attempt to fix a potential build problem.
FWIW, I've locally replaced all occurrences of File.toURL() in the  
codebase to File.toURI().toURL(). Once I've confirmed this breaks no  
tests, I'll commit the changes, so this is out of the way.


Thanks a lot.

Going through the occurrences, I'm getting the impression that in  
some cases, we don't even really need the URL. The URI would do just  
fine if we don't need the functionality for opening a stream... Maybe  
in this particular case, passing through a URL could also be avoided  
(?) If the error is generated by the used StreamResult implementation  
calling File.toURL(), then something like:


Result res = new StreamResult(outputFile.toURI().toASCIIString());

would already be enough (?)


In most cases, yes. But there's really a difference between a URI and a
URL. But what's confusing is the following (the Javadocs for
StreamSource):

quote
public StreamSource(String systemId)
Construct a StreamSource from a URL.
Parameters:
systemId - Must be a String that conforms to the URI syntax.
/quote

URL and URI are both used here. But I think URL is the mandatory term
here. The other thing is URI Syntax which does not refer to URI
itself. Since a URL is a URI, but not all URIs are URLs, I believe your
example above is slightly incorrect.

Jeremias Maerki


It's a SystemId. The class includes setSystemId(...) and 
setPublicId(...), so a URL looks right. One way to find out.


--
Peter B. West http://cv.pbw.id.au/
Folio http://defoe.sourceforge.net/folio/


Re: svn commit: r676161 - /xmlgraphics/fop/trunk/src/java/org/apache/fop/events/model/EventModel.java

2008-07-12 Thread Andreas Delmelle

On Jul 12, 2008, at 14:57, Jeremias Maerki wrote:


On 12.07.2008 14:49:03 Andreas Delmelle wrote:

Going through the occurrences, I'm getting the impression that in
some cases, we don't even really need the URL. The URI would do just
fine if we don't need the functionality for opening a stream... Maybe
in this particular case, passing through a URL could also be avoided
(?) If the error is generated by the used StreamResult implementation
calling File.toURL(), then something like:

Result res = new StreamResult(outputFile.toURI().toASCIIString());

would already be enough (?)


In most cases, yes. But there's really a difference between a URI  
and a

URL. But what's confusing is the following (the Javadocs for
StreamSource):

quote
public StreamSource(String systemId)
Construct a StreamSource from a URL.
Parameters:
systemId - Must be a String that conforms to the URI syntax.
/quote

URL and URI are both used here. But I think URL is the mandatory  
term

here. The other thing is URI Syntax which does not refer to URI
itself. Since a URL is a URI, but not all URIs are URLs, I believe  
your

example above is slightly incorrect.


Nice catch. For the moment, better to be safe than sorry. Indeed, a  
URI could also be something not referring to a file (or, better put:  
a valid input source for a transformation)
Strictly speaking, according to the above definition, the systemId is  
allowed to be a 'mailto:' (?) :-/


Sloppy editors... ;-)

Cheers

Andreas