On Fri, Jul 9, 2010 at 12:20 PM, Daniel Carrera <[email protected]> wrote:

> I'm glad that you like the page. Thanks for your suggestions. I'll add
> a Google search link some time today or tomorrow. Responses below:
>
> On Fri, Jul 9, 2010 at 6:17 PM, P Kishor <[email protected]> wrote:
> > http://pdl.perl.org and http://pdl.sourceforge.net are the same sites,
> > hence, in my view, should have the same URI. Making one or the other
> > the canonical identifier would be very helpful. (I would prefer
> > pdl.perl.org).
>
> Nothing I can do about this. PDL's website is hosted at SourceForge. I
> cannot force pdl.sf.net to *not* work.
>
> > A ubiquitous search field on the web site would be a fantastic help in
> > the above, kinda like an analog of 'help command' at the perldl>
> > prompt. If I am going whole-hog with my wishlist, an auto-suggest for
> > the search field would be very cool, as well as a fuzzy search on the
> > backend. In any case, most site search tools should be able to do the
> > latter. The quickest way to implement a site search might be, however,
> > to implement a Google custom search. It is a few lines of code, and
> > just pops in the page. Then you just wait for Google to index the
> > page, and you are all set.
>
> A Google-based search field is reasonable. An analogue to "perldl>
> help foo" is not. For one, we'd need to have PDL running on the server
> which may not even be possible. Then we'd have to make PDL produce
> help content without the shell, and produce its output in HTML. Ugh.
>

I've been thinking about this myself, since I wrote that functionality into
piddlebot a month or two ago. It's not hard to search the docs database, but
you'd have to have the docs database and modules available to a perl script
on the web server. Of course, you only get the docs database and modules by
installing PDL on said server, and I imagine that is not feasible.

FWIW, the docs database actually has some annoying holes in it that a Google
custom search would not have. Would it be possible to have the custom search
include mailing-list stuff? That would be even more helpful.

 > I figured that I could just change the URI in the address bar. So
> >
> > http://pdl.sourceforge.net/?docs=Core&title=PDL::Core
> >
> > The URI seems to have a possibly superfluous title=<title>. If I just
> > change the docs=<doc name>, I go to the correct page, but sans the
> > title. For example,
> >
> > http://pdl.sourceforge.net/?docs=Ops
> >
> > takes me to the Ops documentation, but now the title is missing. Of
> > course, I could add the title, but that can lead to interesting
> > possibilities as by
> >
> > http://pdl.sourceforge.net/?docs=PDL&title=Scipy
>
> Well, you don't normally navigate by typing GET parameters. And if you
> do, and you type "title=Scipy", I think it's fair to say that the user
> knows exactly what he did.
>
> Anyway, I designed he site the way I did because I don't t want to be
> restricted to using the file name as the page title. Some times I can
> improve clarity by editing the page title. For example:
>
> http://pdl.perl.org/?docs=FAQ&title=Frequently%20Asked%20Questions
> http://pdl.perl.org/?docs=perldl&title=Perldl%20shell
> http://pdl.perl.org/?docs=Impatient&title=Quick%20Start%20Guide
> http://pdl.perl.org/?docs=PP&title=Pre-Processor
>
> And so on.
>

We could have the PHP script that generates the doc pages scrape the
original HTML source to extract the page's title, but I rather admire the
sheer laziness of specifying it as a GET parameter. (I'm not being sarcastic
- I actually like it :-)

David

-- 
Sent via my carrier pigeon.
_______________________________________________
Perldl mailing list
[email protected]
http://mailman.jach.hawaii.edu/mailman/listinfo/perldl

Reply via email to