Hello,
I just noticed that
fileset containting pattern like this "foo*/xxx" are
considered recursive and therefore DirectoryScanner scans whole directory
structure. That is not (completely) true - those patterns could not contain
files in other depth than 1.
I'd like to come with some
optimalization to speed up this case. I tried to add another condition into
"Only include the valid patterns for this
path" part of DirectoryScanner, which would do
some partial-regexp-match to see, if any deeper object could potentially satisfy
that pattern. if not - its not neccessary to scan
deeper.
e.g.
<fileset id="buildfiles" basedir="c:\test\foo"
defaultexcludes="false">
<include name="Gordic*/*.csproj"/>
</fileset>
<include name="Gordic*/*.csproj"/>
</fileset>
c:\test\foo\etc should be discarded when scanned and
not dip into, since none file/folder in it could ever satisfy any
pattern.
c:\test\foo\Gordic1 shouldn't be discarded and scanned
normaly as today.
c:\test\foo\Gordic1\_svn shouldn't be discarded as well
What do you think about this optimalization? Any idea
how to (easily) perform
it?
Martin
