Pixie development has been very interesting lately. Jean Louis Leroy of Tangram fame pointed out that you can change what Data::Dumper uses to serialize objects (ie, use something else instead of bless) which made the clever-but-nasty bless trick go away. Further investigation into Data::Dumper revealed $Data::Dumper::(Toaster|Freezer). We have some fairly good ideas about how to implement locking using them.
pdcawley - brilliant as ever - has fixed Pixie so that there is no longer a need to stuff any metadata inside an object, it just gets magically tracked and handled. It also means we can store array based objects without requiring that the class it is created by implements a get_oid method. This is a Good Thing.
Most importantly Pixie is slowly maturing into something a little more pleasant to work with.
Hooray!
Re:This is good news
james on 2002-08-10T08:24:48
I see an application I would like to build using Pixie to provide persistence for simple user, form and page objects in a CMS. Do you think its ready for use yet?
Yeah, I think it is ready for use. Its got plenty of tests now, and the toplevel API at least hasn't changed, and I can't see it changing. Everything else that comes along will be bells and whistles on top of the basic functionality thats already there.
Things like Indexing and complex queries can wait for the large part, because the ability to name objects can be used to get around the issue.
Another reason why indexing can wait for now, is that the most common thing to do with an object database is to have a batch process run out-of-band and fetch the data from new or changed objects to insert into a RDBMS. At that point you have something you can query against for issues such as reporting.
