Re: [Gnue] reports (clarification)

["Derek A. Neighbors" <derek gnue org>]

> Well, as I've done sometimes, I had a look this morning at the GNUe IRC
> logs, and I saw there some comments from Derek which convinced me to
> write this mail. As I think we're suffering from a big misunderstanding,
> here are some clarifications:

You mean people actually read those logs. :)  Yes as mentioned in IRC I
think a lot of it is communication gap. :)
 
> * and, I thought we had agreed the other day about the general report
> structure, so here is what I understood, please correct me if it's a
> mistake:

me too and thus some of the frustration. :)

> - no one is working on the DTD, this has been abandoned (for the moment,
> since it's not a bad idea to have a DTD AS SOON AS the XML report format
> is finished, but not before of course).

I dont have issues with a DTD or working on one, but I think the guts of
the report engine are far more important than small details in a dtd at
this stage.

> - current work is on making a basic IDL to start writing the report
> engine. As soon as we agree on this, the report engine and the clients
> can start. As you may have noticed, I started very simply to add other
> things as soon as we code and need them.

I would like to see the following even before an IDL or con currently
with an IDL.  

a.  What the template xml file should look like (a small example)
b.  What a the populated report xml file shoudl look like (a small
example)
c.  IDL for pushing the template to server
d.  IDL for processing template (read populate with data)
e.  IDL for specificing what to do with populated file (read convert to
something else or hand back xml to client)

As with these pieces one could begin to author a client and converter at
same time.

> - there will be a XML format to describe a report, that is, not the
> output, but the structure. This format, I guess, could be the one
> produced in the current gda-report.dtd file, since I think it's been
> discussed deeply, although I confess I haven't participated too much on
> that thread, so maybe a new revision by everybody can be made.

I havent even looked, but that would probably do.  A small sample would
be good place to start for proof of concept.

> - this XML format will be processed by the report engine (not the
> client), which will get the data either from GDA (default one) or from
> another source (by means of plugins maybe) to produce a XML output
> format, where the data is actually formatted in another special XML
> format. As we said, gda-report will be made libgda-independent, so that
> anybody could add a different way of accessing its data by just writing
> a plugin. This will be the case for GNUe, isn't it?

Yes this is the case.  Some method for getting a datasource will be used
to merge data with the template into a new xml file.  (How Reinhard and
co wish to go about this is undecided)  libGDA will play a part, its
just a matter of whether bizlogic will sit in front of it (for GNUe).

> - then, clients will ask the report engine for this XML output, or for a
> given format, such as PDF, PS, HTML, etc This conversion, as Derek
> suggested, could be made though a customized jade

Correct.  Jade is a suggestion only.  We could write our own our us XSLT
or the likes if we wish.

> - the report engine will have a sort of cache to return repetitive
> reports, etc

It will have ability to do few things as I see it.

a.  it can cache the converted output... ps, html, pdf etc...
b.  it can cache the xml file.  then a request can be made to do just
the conversion to ps,html,pdf etc...  This could come in handy when you
want to archive something, but may need it in one of several formats. 
If you rarely access it, you save disk space by only archiving xml and
generating output type you need.  Anyhow that is a tangent from this
discussion.

> - apart from this, but outside of the report engine, will be report
> clients and designers, which, of course, won't be started seriously
> until an IDL is approved. This part is, I think, the one Carlos is
> talking about.

I would say designer and client is far less important than a working
engine that is robust.  Its very easy to hand hack the first xml file
template.  Then command line to run through engine and produce
ps,html,pdf is a HUGE step.  As it allows real business stuff to be
done.

I think a client is next most important as it allows then more visual
way to run reports, with a designer being last but definitely not least.
:)
 
> I have not entered technical discussions about each point since this
> will go the next set of mails. But one thing I'm sure about is that GNUe
> and GNOME-DB/libgda should work together on this, it would be a pity to
> have a separation, so please just say what you don't agree and we'll
> discuss it. But really, I thought we had agreed on all this.

Yes I think we should work together.  I thought we had agreed on this as
well, but I saw a lot of client type stuff and dtd stuff being discussed
so I thought I had missed something.  For that I apologize.

The other issue for GNUe is a designer and client that is not xplatform
is dead in the water for us. :(  But that is somewhat of a different
issue.

Derek Neighbors
GNU Enterprise
http://www.gnue.org
derek gnue org