[yaala] feature request

[Florian Forster]

--bgQAstJ9X1Eg13Dy
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Tue, Nov 09, 2004 at 04:50:58PM +0600, qMax wrote:
> here I see/need 2 variants:
> 1. when sorting specified in select with SORT, a-la SQL,

That's what I planned to do

Actually you mostly need two variants: Sorted by field-values and sorted
by aggregation values. The things to consider are: How do we express
something like ``sort field1 by field-values and field2 by aggregation
values''? And, when selecting several aggregation, which aggregation to
use?

In this case I'd suggest the following:
(a) The default sorting is done by field-values
(b) You can select an aggregation with an ``SQL-statement'' along these
    lines:
    SELECT agg1, agg2 BY fld1 ORDER BY agg1, fld2 ORDER BY agg2, fld3;

> 2. SORT not specified, and it make a sence to make sorting for _every_
> aggregation in a table, generate alternative versions of tables for
> each sorting, and generate links to alternative tables in table headers,
> just like in windows widgets.

I didn't think about that, but I think it's better/more flexible to let
the user select what (s)he needs..

> FF> - title for pages
> titles already done - see my previous posts with patches.

I know, that's why it's on the list in the first place ;)

> FF> - n-dimensional tables
> working on it.

Great :)

> >> q> and skipped axes (kinda "BY foo,,bar" or "BY foo,,,bar").

What do you think about this approach: We create a boolean config option
``two-d-tables'' with the default setting ``auto''. Yes/no always/never
create 2D-tables, auto only creates 2D-tables when the horizontal field
has less than, say, 10 values.. This would allow for a compact table if
it makes any sence and prevent huge tables from being generated.

> FF> Since you can only draw 1-d and 2-d tables in HTML you have to
> (i've got an idea about cute 3D... stay online.)

Okay, that'd be interesting ;)

> This is usefull, egg, if I do "select ... BY to,,from" for maillog.
> and i want to make separate 1D-tables with values of 'from' for each valu=
e of 'to'

That's what I meant, I was just a bit unlucky with putting it in english
terms ;)

> FF> Also, could you enlighten me, what exactly your patch for
> FF> ``classic_skip_empty'' does? It didn't make much sense to me this
> FF> weekend..
> This is for 3D tables in Combined -
> to skip empty rows in all of those 2D tables (with val3 set).

Oh, I wasn't aware this could happen at all. I guess it only happends
with 3D reports, right? For n-D reports we'll definitely need this,
since too many useless field-values will be present..

Regards,
-octo
--=20
Florian octo Forster
Hacker in training
GnuPG: 0x91523C3D
http://verplant.org/

--bgQAstJ9X1Eg13Dy
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (GNU/Linux)

iD8DBQFBkKyajvke9ZFSPD0RArUMAJ9g9n+SabAIdD7ypX5GZ0CpFhCvhACglDir
D2V+UTwkF4XX1MgwDWt0ec4=
=v249
-----END PGP SIGNATURE-----

--bgQAstJ9X1Eg13Dy--