Re: [gnue-geas] Re: Proposed changes in libgda

[Rodrigo Moya <rodrigo linuxave net>]

> > Not sure if I explained that well. I understood that the gda-common
> > library provides functions that will be used by the providers as well
> > as by the "clients" of libgda (the code that is on the client side
> > of gda.idl). The report engine is actually a libgda client.
> 
> My understanding that the report engine would not be directly related to
> GDA and would be part of the GNUe project.  There would be a gda report
> client however?  If this is not the case we need to pursue something
> separate in the GNUe project.  Mainly because we need reports to talk to
> GEAS and not libGDA.  
>
Derek,

I've already started the report part in GDA because:
* gnome-db people involved in it did want to have it in GDA. This is not because
  of selfish or ego issues, but because having it outside of GDA would mean
  repeating a lot of code which is already in GDA (mainly OAF/CORBA/GConf).
  And as you can imagine, most of the people involved in this work on this
  on their free time, and trying to convince them to repeat code which is
  already stable and working was quite difficult if not impossible. This is
  the most convincing reason for me, since there is lots of code in GDA which
  will have to be rewritten for the report part to be out of GDA.
* I think I understood that you (GNUe) were going to write a GDA provider for
  communicating with GEAS. That is the schema you said in IRC about
  report->GDA->GDA-GEAS-server->GEAS (or something similar, sorry if it's not
  exactly what you wrote). Having it like this offers all the flexibility
  you want: middleware, DB independence, and just making your report clients
  depending on GDA (libgda-client), which is now, as you know, 100 2.49451NOME
  independent, and which means only depending in a couple of shared libs,
  not on the whole architecture.
* we may want to add XML queries as a second way of executing SQL commands,
  and as it's been already discussed, XML queries MUST go to GDA because
  they'll be used in other parts of the architecture. And if you want total
  DB independence for your reports, I think you should really want to have
  XML queries in the report engine.
* also, we may want in a future to extend the set of DB operations available
  to the report engine. Having the report engine out of GDA will mean that,
  if we ever want to do this, we'll have to recode again some of the code
  present in GDA. I can't show you an example of this right now, but I'm sure
  there will be something really useful to be added this way.

I myself don't see anything that may be painful for GNUe. The only thing that
I can think of is what Derek said about how difficult was to convince people
about having the report engine outside of GNUe. But this is the same as with
any other non-GNUe software you're using right now: ORBit, Jade, etc. Of
course, I may be wrong.

I hope we can reach a concensus on this, but please take into account what
I've said, specially the "repeating code" part, which is IMO, the most
important of all the issues. This is why, although I seemed convinced last time
we talked about this on IRC, I have (again) changed my mind.

cheers