I figured out what I would call a bug.
I'll give an example:
Let's say we have this table:
product = db.define_table('product', Field('cost1'), Field('cost2'),
Field('total', compute = lambda r: r['cost1'] + r['cost2']))
And if you do this:
product.insert(cost1=10, cost2=20)
p = db.product(1)
p.update_record(cost1=5)
This update won't work. What is worse is that it fails silently. The
reason it won't work is that the row input of the lambda does not have
"cost2" in it.
If, however, you do this:
p.update_record(cost1=5, cost2=p.cost2)
Then it will work, because now the parameter r (i.e. the row) has both
cost1 and cost2.
I think it'll be a little tricky for you to fix this, because you probably
want update to have only relevant fields so that saving to the database is
effective. However, this behavior clearly is not desirable.
On Saturday, September 22, 2012 5:23:43 PM UTC-5, Massimo Di Pierro wrote:
>
> Can you show us an example?
>
> computed fields usually do not work when there is not enough information
> to compute. The information must be in the arguments of update_record or
> update. It will not retrive the info from the database.
>
>
>
>
> On Saturday, 22 September 2012 15:43:59 UTC-5, VP wrote:
>>
>> - sqlite
>> - 2 compute fields do not seem to get call on update using both
>> update_record and db().update
>> - They do work on crud/form updates though.
>>
>
--
