[freebase-gridworks-dev] heads-up: row/record model refactoring

8 views
Skip to first unread message

David Huynh

unread,
May 18, 2010, 3:25:19 PM5/18/10
to freebase-gr...@googlegroups.com
Hi all,

I'm planning on layering in a more solid "record" model on top of the
basic "row" model. For example, there is now a RowFilter class, and I'm
planning on adding a RecordFilter class. The FilteredRows class will
also have a corresponding FilteredRecords class. The browsing engine
will have 2 modes: row and record. Facets will be able to count records
properly in the record mode. The "show dependent rows" checkbox will be
replaced by a mode toggle between "row mode" and "record mode". The
table view will handle records much better (it'll show whole records
when in record mode). I think overall this will make Gridworks more
usable. Please let me know if you foresee any problem.

Once this is solid enough, I'll get started on the support for sorting,
which behaves differently appropriately in row and record modes.

I'll also need to add support for changing column groups and key
columns. Once that's done, in the longer term, we can also think about
different data views, such as a tree-like view that's better suited than
the tabular view to show the record model. Maybe the schema skeleton can
be grafted right on top of that.

David

Iain Sproat

unread,
May 18, 2010, 3:30:55 PM5/18/10
to freebase-gr...@googlegroups.com
+1

Stefano Mazzocchi

unread,
May 18, 2010, 3:40:36 PM5/18/10
to freebase-gr...@googlegroups.com, David Huynh
+1

--
Stefano Mazzocchi Application Catalyst
Metaweb Technologies, Inc. ste...@metaweb.com
-------------------------------------------------------------------

Thad Guidry

unread,
May 18, 2010, 3:56:22 PM5/18/10
to freebase-gr...@googlegroups.com

I'll also need to add support for changing column groups and key columns. Once that's done, in the longer term, we can also think about different data views, such as a tree-like view that's better suited than the tabular view to show the record model. Maybe the schema skeleton can be grafted right on top of that.

+1 - that schema skeleton grafting would be right in line with what I saw demo'd for Powercenter, btw.  So great idea !

-Thad 

Thad Guidry

unread,
May 18, 2010, 4:01:43 PM5/18/10
to freebase-gr...@googlegroups.com

I'm planning on layering in a more solid "record" model on top of the basic "row" model. For example, there is now a RowFilter class, and I'm planning on adding a RecordFilter class. The FilteredRows class will also have a corresponding FilteredRecords class. The browsing engine will have 2 modes: row and record. Facets will be able to count records properly in the record mode. The "show dependent rows" checkbox will be replaced by a mode toggle between "row mode" and "record mode". The table view will handle records much better (it'll show whole records when in record mode). I think overall this will make Gridworks more usable. Please let me know if you foresee any problem.

 
When you say on top of the basic "row" model...  That won't go away, but we'll simply have 1 more additional mode for the browsing engine called "record"?

With specific classes built for that "record" mode ?

Will there be specific classes for the "row" mode as well, or not the intent here ?

-Thad

David Huynh

unread,
May 18, 2010, 4:54:08 PM5/18/10
to freebase-gr...@googlegroups.com
All the row stuff will be there and will make up the basic / raw model. A data set consists of an ordered list of rows, each row an ordered list of cells. On top of that raw model we can have zero or more high-level models. One of the high-level models is the record model. The schema alignment skeleton is another model that lets you "see" the same raw data as triples. Different parts of the UI will let you interact with different models. The facets and the main data table view will let you deal with both the raw row model and the record model, and let you switch between them through a UI toggle.

David

Thad Guidry

unread,
May 18, 2010, 5:07:34 PM5/18/10
to freebase-gr...@googlegroups.com
All the row stuff will be there and will make up the basic / raw model. A data set consists of an ordered list of rows, each row an ordered list of cells. On top of that raw model we can have zero or more high-level models. One of the high-level models is the record model. The schema alignment skeleton is another model that lets you "see" the same raw data as triples. Different parts of the UI will let you interact with different models. The facets and the main data table view will let you deal with both the raw row model and the record model, and let you switch between them through a UI toggle.

David


Great!  That's what I was hoping you would respond with.

I understand more completely now, and don't forget to take beautiful pictures of the bugs along the way!

Thanks David,
-Thad
Reply all
Reply to author
Forward
0 new messages