Re: wish: could composite store info about datastores that have been changed?

Marja wrote:

> Axel,
> 
> I got my own HTTP server working and have been able to test more easily.
> And I can update Web files now with HTTP PUT upload. I will do some more 
> testing and cleaning and then make a new preliminary package. I hope ftp 
> will work too with minor corrections but I still need to test that.
> 
> I agree with you that having a RDF server is often better and we are 
> using servers too. Mainly for annotations but we have done experiments 
> with bookmarks as well. However, after having done that I strongly think 
> it is good to also have simple interfaces for normal people who may not 
> have access to RDF servers right away. And being able to update RDF web 
> files gives me just that. So, please don't throw that away.

I'm not throwing it away, because it was never implemented. But I'm 
pretty sure we won't implement it either. Being able to PUT is just 
opening a can of worms, esp from the support front.
I'm not going to open an API that ends up asking me how to configure 
webservers to actually work with that. I'm not sure how that handles 
concurrency issues, too. It *may* work for your particular case, but in 
general, it's flawed.
Note that I'm throwing away the RDF/XML Datasource singleton paradigm as 
well some day, which will make this behaviour even more unpredictable.

>  > All our interfaces are deprecated without replacement. They're not as
>  > good as they should be, and the separation between reading and writing
>  > is one of the things we have on our queue. Tentatively, this is not
>  > really decided upon yet.
> 
> I separated reading, writing, and updating by using several composites, 
> but they mainly help keep track of the files with some of my own 
> tracking. I'll try to do my own wrappings around them as packages when I 
> have time.
> 
> I have not seen anywhere that RDF interfaces are deprecated. If they are 
> it is not clearly communicated to interface users. E.g. if I look from 
> xulplanet there is nothing:
> 
> http://www.xulplanet.com/references/xpcomref/xpcomref1.0/nsIRDFCompositeDataSource.html 
> 
> 
> And the same is true in many other places. Maybe this info could be 
> offered more readily?

The interfaces to RDF have no status at all, so they may be in any 
status. Any interface that is considered stable is actually indicating 
so by having a status. If you don't see it, use it with *lots* of 
precaution, it may change, go incompatible, disappear.

Hopefully with notice, but that depends.

Axel