Re: ~*~ [spam] recommended way of messing things up?

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

Re: ~*~ [spam] recommended way of messing things up?

Dale Henrichs
Norbert,

implement the method in GsSqueakCommon (don't proxy for Grease) and add a test or two ... Grease is now part of the base install (because of the XML stuff:) so there are potential dependency problems ... submit an issue on glassdb noting the duplication of methods and I'll clean up Grease at a later date ...

Note that I am in the middle of development of GLASS 1.0-beta.8.7 so I have development versions for all of the configs open and aimed 1.0-beta.8.7, so if you are inclined to snap off a version of the GsSqueakCommon config, just create a point release (branch off of the 1.0-beta.8.6 version) and I'll merge your changes at a later date ..

I've pushed the stack on GLASS for the moment and am diving into the seaside upgrade for 3.0 and am focussing heavily on this ...

Dale

On Mar 29, 2011, at 11:03 AM, Norbert Hartl wrote:

> While upgrading the xml parser I struggled with XMLWriter that uses
>
> WriteStream>>next:putAll:startingAt:
>
> that is not available in gemstone. The method exists in squeak and in pharo. And....in grease as greaseNext:putAll:startingAt: So I'm a bit insecure what is the best way to fix it. Well, at first it didn't notice the grease version and implemented it in GsSqueakCommon.
>
> I think the story goes like this. XMLWriter uses this method and there is no easy replacement that I've found. XMLWriter does not depend on grease so no way using this instead. So I have the options:
>
> 1. Fork the XMLWriter package for gemstone to fix this. This I don't like at all because forking is a big side effect
> 2. Implement the method in squeak common. That would be the dumbest and most safe on by copying the code.
> 3. Implement the method in WriteStream that delegates to the grease version with method category *gssqueakcommon-core. That really depends on the package structure in gemstone. This would be the least code doubling but the biggest danger to mess something up
>
> What would be your choice? Is there another alternative?
>
> Norbert