> > It appears to work pretty well, but it would be nice to see a bit more
> > structured support of this, although, to be frank, I am not sure what that
> > would look like. Although I do know one thing I would like, and that is
> > more of a programmatic API and linkable library into htsearch, not just a
>
>Many people have mentioned this--it's certainly a good idea. I imagine as
>a 3.2 release firms up, the class interfaces can be finalized. But
>feedback is certainly welcome as to "what it would look like."
Hmm. I'm not sure if you are agreeing that support to separate logic from
display is a good idea, or a programmable interface. Assuming the latter ...
I compared SWISH-E, SWISH++, and htDig about two months back. I eventually
implemented htdig as I like htDig's support for excerpting. But I
remember thinking that the SWISH "family" had better programming APIs
(although most has reputations for being pretty buggy.)
I think a library I would like to see would have interfaces that let me:
1. get a handle to a index
2. specify a search and get a handle to search results back
3. obtain next <n> results (where n is a parameter, and
you loop until no more results or until you close the search handle
4. close search handle
5. close index handle
Regarding 3, I can imagine that the output is either/both
a. A list of strings (following the long/brief/... template) that describe
the e next n results
b. a list of specific result handles that you then have another interface
that lets you get information back about that result
resultInformation *htSearchOneResult(handle)
I am lisp, tcl and xml biased, but I would suggest not creating these
things to take or return complex C++ structs, but to take/return arrays or
lists of simple string based descriptors (and simple xml descriptions would
be wonderful.) The goal would be to make it trivial to create efficient
interfaces to the htSearch library in C, C++, C#, perl, Tcl, Python, Java,
whatever...)
Regarding specification of display and results, I guess I am pretty happy
with the template_map mechanism, although I do think it would be good just
to natively support an XML description as well.
I am willing to help out, within my time constraints, and within my
ignorance of C++ and htdig internals....
Thanks,
Jerry
_______________________________________________
htdig-dev mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/htdig-dev
