Re: tolog for TMQL

On Fri, Apr 08, 2011 at 10:39:02PM +0200, Lars Marius Garshol wrote:

> > [ BTW: I cannot remember having seen subquery functionality in tolog. ]
> 
> It's not there. It probably should be. 
> 
> I see no major obstacles to adding it. 

See below. [A]

> > The issue is - as always - the _locality_ of the information. Here I
> > still contend that PEs - by definition - have a narrower/constrained
> > way to pass on information.

> If your point is that tolog requires a global analysis of the query
> to tell whether variables are used elsewhere or can be safely
> optimized away then that's certainly true.

Not exactly my point: In TMQL@rho

   ALL variables

are optimized away. Always.

So during evaluation there is never a need to refer to any
variable[B]. All what is necessary to evaluate a part of the PE is
what is "coming from the left" (and the context map, of
course). Consequently, a PE can run on different nodes, as long as
these have access to the same map.

I found that feature of TMQL@rho cute.

Adding subqueries [A] in TMQL@rho does not change that. I cannot say
how this pans out in tolog@larsbot.

\rho

[B]: Not precisely true: All "variables" which are passed into the
query from the application should actually be converted into constants
during evaluation time. I still do not do that.