[ts-gen] QQQQ example: was: Re: Some Newbie Questions

[R P Herrold]

On Tue, 15 Jul 2008, Jay Strauss wrote:

> Hi Russ,

;)

> Thanks for the guidance.  So you've answered a bunch of my 
> questions so far:

> 1) QQQQ (the ETF) is already in the DB, so it's the 
> developer's responsibility to query the DB to find the ID to 
> submit to the shim.

This will change shortly (get much easier), but, I wanted to 
get your question answered as you were plowing through 
(ploughing thru) the details.

> 2) The DBs preloaded data is maintained in general by you, 
> but with suggestions from the Mailing list.

yup -- again, Bill and I were working on this, as you raised 
the topic, in our planning meeting yesterday (we do them on 
Mondays), and again today, and this will only get easier in 
the future.  There is an irreduceable minimum however, which 
we will reach, at which point, database unloads and reloads 
will become needful.  We have tried to finesse this during 
the developmental phase [where we have needed the freedom to 
alter the schema to repsond to un-anticipated coding 
implementation details], but we know it is on the horizon.

> 3) "Select" is kinda new and doesn't show in the docs, I 
> should use the source (luke), as my guide.

I think the sample scripts in ./exs/ might be easier, along 
with my 'commands.pdf' referece today;  the source is ... not 
for the timid, being what I consider a very mature codeing 
approach in C++.

> 4) With regard to Oracle.  Oracle has a separate device/ 
> entity/ construct/ table... call a "sequence" which one uses 
> to generate "ID"s for tables.  For each table, one 
> (typically) creates a corresponding sequence, and then link 
> the two with a "on insert" trigger.  Sequences are 
> guaranteed to be unique across login sessions so there is no 
> risk of duplication.  I know lots of other RDBMS have an 
> "auto increment" type of column.  Its a different design, 
> both have advantages and disadvantages, but they both 
> accomplish the same goal.

Got it.  Thanks for clearing that up.  I started writing what 
we now call relational databases back in the early '70's, and 
know that there are many roads to wrangling data  ;)

> 5) regarding the shim output being a file, pipe, syslog.  I get it.

Only think I would add is that the shim can (trivially) do 
several at once in its output routing.

-- Russ