[Erp5-dev] start_date on non-movement object

[bartek]

Łukasz Nowak wrote:
> Hello,
> 
> On 2008-03-25, 13:14:33
> Łukasz Nowak <lukasz.nowak at ventis.com.pl> wrote:
> 
> (...)
>> Using revision 18850. Thanks for pointing out this testing
>> scenario - more things become clear.
> 
> And taht is my fault. After reading KnownBugs from "common Polish wiki",

This is our internal document which refers to r18850, I don't know what 
is the HEAD status here.

Anyway, it is strange that it solved your problem; the sqlvar.py issue 
causes activities to be fired one hour too late - but apparently only in 
Poland :) I've discussed the issue with Ivan and he said it works in all 
timezones.

B.

> right now all is working correctly, with such patch:
> Index: sqlvar.py
> ===================================================================
> --- sqlvar.py   (revision 18850)
> +++ sqlvar.py   (working copy)
> @@ -71,10 +71,11 @@
> 
>          try:
>              if getattr(v, 'ISO', None) is not None:
> -                v=v.toZone('UTC').ISO()
> +                v=v.ISO()
> +            elif getattr(v, 'strftime', None) is not None:
> +                v=v.strftime('%Y-%m-%d %H:%M:%S')
>              else:
> -                v = DateTime(v)
> -                v=v.toZone('UTC').ISO()
> +                v=str(v)
>          except:
>              if not v and args.has_key('optional') and args['optional']:
>                  return 'null'
> 
> Sorry again. That was my fault. I'm not good enough in reading unit
> tests, and haven't properly noted about this bug without bugtracker.
> 
> Problem solved^Wworkarounded (no time to upgrade to HEAD from 18850
> to check if on clean will it work).
> 
> Regards,
> Luke
> 


-- 
"feelings affect productivity. (...) unhappy people write worse 
software, and less of it."
Karl Fogel, "Producing Open Source Software"