Hi,
I am looking at extending the xslt task to allow for an enhanced check
to determine whether a transform should proceed. Currently there are 2
options for checking whether a transform should proceed. The first
just checks the XML and XSL Source against the output file. The 2cnd
uses a DependSet. While the DependSet can check the transform at a
deeper level, it is inconvenient to set up. Also, in many ways it is
impractical, because, for example, a directory containing src XML
files could have each reference their own set of files (through the
document function).
We discussed this a bit on the user list here:
http://marc.info/?t=122418056700022&r=1&w=2
Specifically:
http://marc.info/?l=ant-user&m=122418293609849&w=2
Basically, I want to use 2 URIResolvers instances (of
CachingURIResolver ?) in addition to the the checks for the primary
XML and XSL Source:
1. Set one on the TransformerFactory to resolve import/include. If
this check turns out to not be up to date, proceed with all transforms
skipping the next check.
2. Set another one on the Transformer to resolve document(). If the
factories test is up to date, then check the dependencies accumulated
through this resolver.
These resolvers would gather up their dependencies to track for
subsequent transforms.
Some questions:
* where is the DependSet being checked in the transformation(s)? I
can't find it :)
* I will need to write to the file system to maintain the cache. What
would be a good way of handling this? Should I create a directory in
the baseDir for all transforms. Then a directory per xslt task. Inside
that dir create a properties file per transform? How would you do it?
* new class name TraxCachingLiason -- ideas?
best,
-Rob
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
