On Wed, Oct 31, 2012 at 12:33 AM, hakan <hexit34@users.sourceforge.net> wrote:
 
I like the rs.i.id style, I haven't used stems as output from oosqlite exec at all, so no code to change here either.
for me it's more OO_STEM_OF_ROWS (rs.0) than OO_STEM_OF_COLUMNS

Yes, I like that style better.  It is really what I first intended to implement, and what I prefer.  Rather than change OO_STEM_OF_STEMS, I just kept it as is and added a fourth format.

Right now it is OO_CLASSIC_STEM.  I can easily change the name, but that seems adequate to me.  Unless there are some vocal opponents to that, I'll just keep it.


 

Regarding the risk of using a local variable with the same name as a table column,
I think that's the programmer's responsibility to keep track of. That problem is always present regardless oosqlite.
So please no construct like, stem.0column or .!column

Yes, I don't like those constructs myself, so I haven't implemented that change.  Again, unless there are some vocal advocates for that construct, I'm not inclined to make a change here.

 

Another thing to maybe (I am not sure) consider is if a table column name have embedded space, like 'my column'.

Is that possible in SQL?  I certainly don't know much about the details here, but I was assuming a column name could not include spaces.

Also if using stem.~allIndexes/makearray, you get an array of all stem tails/indexes, but it's in an unspecified order and in the case that
this is table column names, it had been nice to have the same order as the sql select statement order.

Currently if you use:  do i = 1 to n you get the same order as the SQL select produces the order.  Not sure if this is just a statement of fact by you, or you have some meaning here I don't understand.  ;-)

--
Mark Miesfeld