[Serusers] Re: [Serdev] New search-page

[Marcin Pazdro]

2007/3/27, Jan Janak <jan at iptel.org>:
> Greger V. Teigre wrote:
> > Yes, this looks good.
> > I agree with Jan that we maybe can simplify a bit. Here are some
> > additional suggestions that you might want to consider:
> > - Instead of having three panes, just link to the documentation. I think
> > this makes it easier to get the iptel.org menus and context. People can
> > decide themselves if they want to open a new tab, window or whatever
> > using shortcuts in their browsers
>
>   Yes, I agree. Simplicity is the key here.
I agree too.
Originally we wanted to make search-page as sexy as possible, so,
documentation loaded on-the-fly into documentation frame was great
idea, but if we go to simplicity...
So there will be just search frames and the generated table with links
>
> > - Maybe name of entity and free text search could be merged into one?
> > I'm not sure if we want to search across anything else than the name of
> > function or param?
>
>    No, search by name and version you are insterested in would be
>    sufficient.

OK, no problem

> > - I think we can start with a web 2.0 app and then see if it makes sense
> > to create something else or maybe just save static outputs that people
> > can view
>
>   What is web 2.0? Do you mean automatic updates through javascript? Do
>   you think this is really necessary for something that simple?

So, as I said. I wanted to create a fully dynamic working search-page
where searching documentation, sorting and loading documentation does
not require page reloading. AJAX! This framework (combination of XML,
javascrip and dhtml is supported only by modern browsers). Originally
it was meant not to be so simple as you think, Jan.

> > - I don't think we currently have any marks (#ref) in the html output of
> > the module docs. But I assume the stylesheet can be adapted. Jan, is
> > this possible?
>
>   Yes, that's easy to do.
:)

> > - I don't think the app will be able to know if the documentation is
> > missing. Only when following the link, the user will be brought to
> > either just to the module doc or, with #refs, to the right location in
> > module docs, and the user will discover that something is missing.
> > So, I think we need a "Report missing or insufficient
> > documentation"-link for each element in the table. Maybe that can be
> > done by creating a form out of the table, a checkbox at the beginning of
> > each row, a comment field at the bottom, and a "Submit missing or
> > insuffient documentation report" button at the bottom

This is a good idea Greger, however...
When a user clicks the link in order to load (missing) documentation I
doubt that after he/she discovers that there is no information he/she
is looking she /he will like to go back to report missing
documentation. But, I may be wrong.
Nevertheless, I am going to implement this as you proposed :)

> > - And finally, I disagree with Jan on the drupal integration. Maybe we
> > will want to include searchable html docs as part of an install package,
> > on another website, or we may (heavens no) switch system. I think we
> > will be able to get the headers, menus etc the same way we have made the
> > module docs (by embedding code as content on a drupal page). Integration
> > is the root of all evil and the more independent we can make this, the
> > better, IMHO

I was playing with all these drupal, taxonomies, hierarchies,
categories... whatever.
Although it seems to be really powerful, I'm afraid that getting
understand how this really works and how to create php applets that
could be easily included to the durpal schema may take much longer
than just to implement new search-page.
So, I consent to Greger's suggestion

>
>   Well, OK, then at least a php function which will return to the
>   documentation for a particular function name and version would be
>   nice.
OK. But we can still make it "sexy" and easy.
We can still use AJAX (Spry framework).
for searching for functions etc.. we can for example get similar effect as here:
http://labs.adobe.com/technologies/spry/samples/data_region/NonDestructiveFilterSample.html
On-the-fly table searching and sorting would not be a problem as well.
More over, we could get rid of PHP at all. It would be enough with
html and XML file. That file should have following structure:
<documentation>
  <!-- Table iptel -->
    <iptel>
        <module>tls</module>
        <version>ser 0.10.99-dev66-tls (i386/linux)</version>
        <name>@tls</name>
        <type>3</type>
        <params>0</params>
        <link>somelink</link>
    </iptel>
    <iptel>
        <module>tls</module>
        <version>ser 0.10.99-dev66-tls (i386/linux)</version>
        <name>@tls.version</name>
        <type>3</type>
        <params>0</params>
        <link>somelink</link>
    </iptel>
    ...
   ...
</documentation>
This file could be generated together with documentation (is that
possible Jan?) or if not, I could create some semi-exe php script that
will generate that XML document once a day.
Avoiding performing PHP scripts, searching databases or generating
scripts as the documentation are requested will save server resources.
:)
Moreover, adding pure HTML (with some JavaScript) would not be a problem
Best regards,
-- 
Marcin Pazdro