rfe: make QueryParser a public class
|
View:
New views
5 Messages
—
Rating Filter:
Alert me
|
 |
rfe: make QueryParser a public class
by Michael Allman
::
Rate this Message:
The subject pretty much says it all.
I want to implement some custom XQuery optimizations. The approach I'm pursuing is using a custom QueryParser and some custom Expressions to build an optimized XQueryExpression. Only thing foiling my plan is the fact that QueryParser is package-private. Currently, I'm creating my own subclass of ExpressionParser which is 99% identical to QueryParser. Obviously a hack. Ironically, the methods I want to override (so far) are all in ExpressionParser. Is this a reasonable/feasible request? Is there another way I should be pursuing custom optimizations? Cheers, Michael _______________________________________________ saxon-help mailing list saxon-help@... https://lists.sourceforge.net/lists/listinfo/saxon-help |
|
|
Re: rfe: make QueryParser a public class
by Michael Kay
::
Rate this Message:
I think there are two levels at which Saxon is extensible. It's extensible
through defined APIs designed for the purpose, such as the extension function binding mechanism, the TraceListener, and so on; and it's extensible by virtue of the fact that the code is open source. At the moment I don't think it's a realistic proposition to provide a supported API for extensibility at the kind of level you are envisaging. So rather than tweaking a few classes and methods to make things public, I think it's better to acknowledge that the kind of thing you are doing requires source code changes. (Actually, there's a benefit to this: under the Mozilla licensing rules, this means you have to publish your changes...) Michael Kay http://www.saxonica.com/ > -----Original Message----- > From: saxon-help-bounces@... > [mailto:saxon-help-bounces@...] On Behalf > Of Michael Allman > Sent: 10 June 2006 02:39 > To: saxon-help@... > Subject: [saxon] rfe: make QueryParser a public class > > The subject pretty much says it all. > > I want to implement some custom XQuery optimizations. The > approach I'm pursuing is using a custom QueryParser and some > custom Expressions to build an optimized XQueryExpression. > Only thing foiling my plan is the fact that QueryParser is > package-private. > > Currently, I'm creating my own subclass of ExpressionParser > which is 99% identical to QueryParser. Obviously a hack. > Ironically, the methods I want to override (so far) are all > in ExpressionParser. > > Is this a reasonable/feasible request? > > Is there another way I should be pursuing custom optimizations? > > Cheers, > > Michael > > > _______________________________________________ > saxon-help mailing list > saxon-help@... > https://lists.sourceforge.net/lists/listinfo/saxon-help _______________________________________________ saxon-help mailing list saxon-help@... https://lists.sourceforge.net/lists/listinfo/saxon-help |
|
|
Re: rfe: make QueryParser a public class
by Michael Allman
::
Rate this Message:
Damn. I was afraid you were going to say that.
FWIW, my intention from the start has been to publish my work as open source under a BSD style license. I just don't like the idea of maintaining my own set of patches against Saxon. That being said, I respect and understand your position. At some point, you might find it worthwhile to open up the parsing/evaluation part of Saxon a little more. BTW, one of the reasons I'm adopting Saxon for XQuery implementation is precisely because it's open source. Thanks for that. Now if only Saxon-SA were open, that would be awesome . . . I'm wondering if my source patches will be compatible with Saxon-SA. Time will tell. I'd like to support Saxon-SA as a pluggable alternative to Saxon-B in my extension. Cheers, Michael On Sat, 10 Jun 2006, Michael Kay wrote: > I think there are two levels at which Saxon is extensible. It's extensible > through defined APIs designed for the purpose, such as the extension > function binding mechanism, the TraceListener, and so on; and it's > extensible by virtue of the fact that the code is open source. At the moment > I don't think it's a realistic proposition to provide a supported API for > extensibility at the kind of level you are envisaging. So rather than > tweaking a few classes and methods to make things public, I think it's > better to acknowledge that the kind of thing you are doing requires source > code changes. > > (Actually, there's a benefit to this: under the Mozilla licensing rules, > this means you have to publish your changes...) > > Michael Kay > http://www.saxonica.com/ > >> -----Original Message----- >> From: saxon-help-bounces@... >> [mailto:saxon-help-bounces@...] On Behalf >> Of Michael Allman >> Sent: 10 June 2006 02:39 >> To: saxon-help@... >> Subject: [saxon] rfe: make QueryParser a public class >> >> The subject pretty much says it all. >> >> I want to implement some custom XQuery optimizations. The >> approach I'm pursuing is using a custom QueryParser and some >> custom Expressions to build an optimized XQueryExpression. >> Only thing foiling my plan is the fact that QueryParser is >> package-private. >> >> Currently, I'm creating my own subclass of ExpressionParser >> which is 99% identical to QueryParser. Obviously a hack. >> Ironically, the methods I want to override (so far) are all >> in ExpressionParser. >> >> Is this a reasonable/feasible request? >> >> Is there another way I should be pursuing custom optimizations? >> >> Cheers, >> >> Michael >> >> >> _______________________________________________ >> saxon-help mailing list >> saxon-help@... >> https://lists.sourceforge.net/lists/listinfo/saxon-help > > > > _______________________________________________ > saxon-help mailing list > saxon-help@... > https://lists.sourceforge.net/lists/listinfo/saxon-help > _______________________________________________ saxon-help mailing list saxon-help@... https://lists.sourceforge.net/lists/listinfo/saxon-help |
|
|
Re: rfe: make QueryParser a public class
by Michael Kay
::
Rate this Message:
> That being said, I respect and understand your position. At
> some point, you might find it worthwhile to open up the > parsing/evaluation part of Saxon a little more. Yes: I'd like to have some kind of API that provides access to the expression tree. One way of doing this might be through a model like XQueryX. But it's not really very high on my list of priorities (a higher priority, for example, is creating a cleaner interface between Saxon and the compiled schema so that I can work with other products' schema compilers, and other products can work with mine). Michael Kay http://www.saxonica.com/ _______________________________________________ saxon-help mailing list saxon-help@... https://lists.sourceforge.net/lists/listinfo/saxon-help |
|
|
Re: rfe: make QueryParser a public class
by Bugzilla from frans.englich@telia.com
::
Rate this Message:
On Saturday 10 June 2006 18:44, Michael Allman wrote:
> Damn. I was afraid you were going to say that. > > FWIW, my intention from the start has been to publish my work as open > source under a BSD style license. > > I just don't like the idea of maintaining my own set of patches against > Saxon. Another solution could be to simply integrate your solutions into Saxon itself. That was what I first was thinking, since optimizations on the level you seem to do easily gain others. Of course, that would likely mean Michael having to spend time on it, so in several ways this might not work out. Cheers, Frans _______________________________________________ saxon-help mailing list saxon-help@... https://lists.sourceforge.net/lists/listinfo/saxon-help |
| Free embeddable forum powered by Nabble | Forum Help |
