Strigi Analyzers

[George Goldberg]

Oh well! :-) ...

There's no point in us wasting effort doing the same thing... Should I
continue with the Cpp one, with a view to adding indexing of class
names, included files etc, or are you already there with it?

Also, I posted a cpp analyzer as a throughanalyzer a couple of days
ago, and someone suggested that it would be better as a line analyzer,
as that option is now available.

George

On 3/21/07, Tobias G. Pfeiffer <tgpfeiffer at web.de> wrote:
> Hi!
>
> On Tuesday 20 March 2007, 20:42, George Goldberg wrote:
> > I'm working on porting the KFile plugins for diffs and cpp files to
> > StreamLineAnalyzers, but I don't know whether they should be
> > implemented as part of strigi or as plugins.
>
> Funny thing, I just started... porting the KFile plugin for cpp :-) However, I
> used the StreamThroughAnalyzer for that, mostly because there were templates
> and docs around for that (and I thought noone else would have the same idea).
> On #kde-devel, Jos van den Oever said:
>
> <vandenoever> in any case write a filename with ....analyzer.cpp
> <vandenoever> and where to put it depends on the type
> <vandenoever> if you need kdelibs or something, then leave it where it is
> <vandenoever> so that it get installed along with the right dependencies
>
> Being really new to KDE development, I think that for now, just create a
> *[through|end|line]analyzer.[h|cpp] file out of the corresponding kfile_*
> thingy and have it remain in that same directory, if that is what you mean.
> If, at some point, it is decided to move all those stream analyzers over to
> strigi, it will probably be easier this way than moving them back into their
> original positions.
>
> > Finally, I'm thinking of
> > writing PHP and Perl script file analyzers in a similar vein to the
> > Cpp file one. Would this be worthwhile. Any answers much appreciated.
>
> Well, I consider this cpp-meta-info like "how many lines of comments" not
> really interesting, neither in C++, nor in PHP or Perl. If you find something
> in these files that's worth being indexed (like names of included files,
> defined class names etc.) and are able to extract this information, then do
> so, I think.
>
> Bye
> Tobias
>
> --
> ...and justice for all!
>
>