atom feed9 messages in com.mulberrytech.lists.xsl-listRe: [xsl] Saxon for C/PHP/Python/etc
FromSent OnAttachments
Dr O'Neil DelprattSep 27, 2013 3:50 am 
Adam RetterSep 27, 2013 3:57 am 
Wolfgang LaunSep 27, 2013 5:23 am 
Adam RetterSep 27, 2013 5:38 am 
Tony GrahamSep 27, 2013 6:07 am 
Adam RetterSep 27, 2013 6:26 am 
Steve BallOct 3, 2013 6:06 pm 
Adam RetterOct 3, 2013 6:41 pm 
Michael KayOct 4, 2013 6:34 am 
Subject:Re: [xsl] Saxon for C/PHP/Python/etc
From:Wolfgang Laun (wolf@gmail.com)
Date:Sep 27, 2013 5:23:12 am
List:com.mulberrytech.lists.xsl-list

What about a JNI interface for the API provided by Saxon? This would avoid dependencies on any additional translator. -W

On 27/09/2013, Dr O'Neil Delpratt <one@saxonica.com> wrote:

Dear XSLT community,

I have recently been looking at the possbilities of making XSLT 2.0/3.0 processor available to the C/C++ world. At present there seems to be a real shortage or a lack of support for anything greater than XSLT 1.0 (i.e libxslt, Xalan, etc).

The benefits are clearly to be seen: having a fullfledged XSLT 2.0 in C would be great for the PHP/Python/Ruby/... communities, who currently rely on libxslt.

So I have been looking at tools that can convert Saxon to native machine code. Namely LLVM and GCJ.

I have had much better success using GCJ given that I managed to compile Saxon-HE to native machine code and actually get it to execute some stylesheets without any problems. However is GCJ the right technology to use for the task of converting Java code to native machine code, since it is somewhat obsolete?

The LLVM project seems much more active, but I am finding it difficult to get anything working. I have tried the VMKit which relies on LLVM to compile some helloworld examples to machine code, but even that seems cumbersome.

I would be grateful for feedback from anyone with experience of these technologies.

Kind regards,

O'Neil