RE: [topicmaps-comment] mapping topic maps on a relation database

[Danny Ayers <danny666@virgilio.it>]

Looking good, and it's a very good sign that the binding will work pretty
much the same Java/Javascript/Python.

Coming from a Java angle at least, I certainly do think that specifying APIs
as interfaces make a lot of sense - abstract classes can make unnecessary
demands on how to implement, especially with the multiple inheritance thing
(subclass or use composition?).

A couple of minor points, but before things get set in stone - the naming is
a little inconsistent e.g. :

getAssocIndex()
getAssociations()

also the typing isn't clear (mainly due to it being JS - definitely Java
interface time), an extreme example : getResourceOrData() -  does it always
return the String "maybe"?

I'd suggest having a looking at the SVG spec appendices - there's a binding
for Java & one for JavaScript that are essentially the same, but the style
in which each is expressed is interesting.

Anyway, apart from niggles, it's looking very good. I've been looking
forward (when time permits) to having a play with TM code, but the prospect
of coding the API from the specs didn't appeal, so thanks guys!

Any chance of making the utility code available? I for one would certainly
be interested.

Cheers,
Danny.

---
Danny Ayers
<stuff> http://www.isacat.net </stuff>


>-----Original Message-----
>From: Thomas B. Passin [mailto:tpassin@comcast.net]
>Sent: 21 April 2002 21:50
>To: topicmaps-comment@lists.oasis-open.org
>Subject: Re: [topicmaps-comment] mapping topic maps on a relation
>database
>
>
>[Lars Marius Garshol]
>>
>> * Thomas B. Passin
>> |
>> | Oh, good, you've got something going on an API.  Wish I had known...
>> | We had a thread on this some months ago and it didn't seem like it
>> | was going anywhere.
>>
>> Actually, it's been around for a while. It hasn't really been
>> announced, though. I know Kal was planning to do this at some point,
>> so I hope I haven't stolen his thunder now.
>>
>> | I won't have time to contribute until late summer at least, I'm
>> | sorry to say, but I'd be glad to let you see the object API I use
>> | with my javascript engine.  It could easily be translated to Python
>> | or java.  This API is targeted towards end applications like
>> | browsers.  Let me know if you would be interested.
>>
>> I would definitely be interested.
>>
>
>Lars has (off-line) asked me to make this API publicly available.
>I agreed,
>so here it is in the attachment.
>
>I have found that this set of APIs is very practical for the applications I
>have built so far.  In fact, I developed it so I could experiment with UIs
>for topic maps.    In addition to the core API I have attached, I have a
>bunch of utility functions, but I don't know that they belong in
>the API per
>se, and haven't included them.
>
>I won't pretend that this API is complete, but it has been a great help to
>me in exploring UI issues for topic maps.  With the help of utility
>functions, you can save maps in a javascript or xtm format, load the apps
>using a javascript serialization (which you can get from an xtm
>map using an
>xslt stylesheet), filter by scopes, and a wide range of other
>useful things.
>It probably has about 80% of the features needed for general apps.
>
>Porting this API to Python from javascript is almost a no-brainer.
> I did it
>in one long evening.  For java, since I use multiple inheritance, you would
>want to make the abstract classes pure interfaces instead.  Then you could
>implement the multiple interfaces where I'm using multiple inheritance.
>
>What's that, you say?  Javascript does not have multiple
>inheritance?  Well,
>it turns out you can simulate multiple inheritance.  Javascript doesn't
>really have inheritance anyway, it uses prototypes in a way almost exactly
>the same way as Python.  Python actually makes a good prototyping language
>for javascript, although in this case I've done it the other way round.
>
>Enjoy!
>
>Tom P
>
>