I was surprised today when I saw this in one of my table entries:
<gs:header row='1' />
<gs:data insertionMode='overwrite' numRows='10' startRow='2'>
<!-- more columns omitted -->
<gs:column index='A' *name=''*/>
</gs:data>
A couple of days ago, the same entry had been defined with:
<gs:header row='1' />
<gs:data insertionMode='overwrite' numRows='10' startRow='2'>
<!-- more columns omitted -->
<gs:column index='A' *name='ID'*/>
</gs:data>
(note that the name has changed from 'ID' to '')
On a whim, I open up the worksheet, and sure enough, whatever I change cell
A1 to is what turns up as the column name in the tables feed.
This isn't the behaviour I expected. Since I:
a) am forced to supply the <gs:data/> section when I
b) am forced to rely on column names when providing the 'sq' query
parameter on the records feed
therefore, I rely on column names *remaining as I defined them when I
inserted the table entry into the feed*. The names I choose to assign to
the columns are my special, table-specific names for those columns which
happen to suit my needs for working with the records in those tables.
In other news, there's also a <gs:options allCols='false' allRows='false'/>
in the feed, which isn't specified in the Spreadsheets 3.0 reference guide
or the RNG schema for the table feed at
http://code.google.com/apis/spreadsheets/data/3.0/schema/table_atom.rnc
Can you guys like, refrain from making *unannounced* changes to the
semantics of the feeds?
Now, can we please sit down like server and client and nail down the actual
contract specified by the table feed? The SS 3.0 guide and reference are
quite inadequate on the significance of the <gs:header/> versus the need to
specify <gs:columns/> (one of these must necessarily be ignored by the
server, yet they are both mandatory in the <entry/>). Please explain what
*should* happen, because I can't actually determine it from the docs.
Amazed in the wrong kind of way,
David.
