status report.
1) I reverted from 'repr' to 'str'. we are going to have rounding
errors with sqlite but not as big as in my example before.
2) I also have added support for decimal for all the other databases
that support it. There is an issue with some adapters so I do not
swear by this. Moreover I did not have time to check it carefully.
I am caching a flight tomorrow morning so I will probably be unable to
look at this again for 24hrs.
If you get a chance to test it please let me know. Just knowing that
you can get data in and out using any database other than sqlite would
be important.
Mind that you cannot do:
db.define_table('a',Field('b','decimal(10,3)))
db.a.insert(b=3.14)
You have to do
db.a.insert(b='3.14')
Massimo
P.S. if anybody from the list is going to be to SuperComputing Sunday-
Tuesday and want to meet for a coffee, send me an email.
On Nov 14, 12:04 am, mdipierro <[email protected]> wrote:
> decimal is a big can of worms:
>
> http://www.mail-archive.com/[email protected]/msg89642.html
>
> On Nov 13, 11:43 pm, mdipierro <[email protected]> wrote:
>
> > Fascinating. As an experiment I replaced str with repr and it usually
> > works with float. Float is better then integer because allows
> > migrations from one decimal type to float and vice versa and allows
> > for expressions involving different field types without major errors.
>
> > The problem is that when it fails, it has major failures.
>
> > Everybody should be aware that there will be much bigger rounding
> > errors with type='decimal' on sqlite than with type='double'. With
> > double the rounding errors will be of the order of required decimal
> > precision, much bigger than double precision.
>
> > That is because is
>
> > >>> a = 0.3/3 # is float
> > >>> print str(a)[0:5]
>
> > 0.1
>
> > while
>
> > >>> from decimal import Decimal
> > >>> b=Decimal(repr(a)[0:5])
> > >>> print b
>
> > 0.099
>
> > Conversions like the one above are necessary step to get data in/out
> > of database.
>
> > The more I look into this the more I think we should not support it if
> > the database does not support it.
>
> > Massimo
>
> > On Nov 13, 9:27 pm, Jonathan Lundell <[email protected]> wrote:
>
> > > On Nov 13, 2009, at 4:36 PM, Massimo Di Pierro wrote:
>
> > > > In theory double has 53 bits. Yet I tried and it does not seem to work
> > > > properly. It rounds to the cents. You may want to try the attached
> > > > files. Perhaps I am doing something wrong.
>
> > > I'm not advocating floating point; I think it's a pain, and if integer is
> > > working, that's better. But....
>
> > > >>> from decimal import *
> > > >>> b = Decimal("1234567891.987654321")
> > > >>> b
>
> > > Decimal('1234567891.987654321')>>> float(b)
> > > 1234567891.9876542
> > > >>> str(float(b))
> > > '1234567891.99'
> > > >>> repr(float(b))
>
> > > '1234567891.9876542'
>
> > > I haven't checked, but it looks like an artifact of float.str.
>
>
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"web2py-users" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at
http://groups.google.com/group/web2py?hl=en
-~----------~----~----~----~------~----~------~--~---
