price.date, transaction.post_date and neutral time

[Wm]

On 13/02/2018 18:12, Sébastien de Menten wrote:
> r/2012/2018/ (it was a typo)

ok

> My point is that a price entered via the price editor (manually) is handled
> differently than a price generated via a transaction 

Isn't that a good thing ? I shouldn't have to say again that the price 
db is just for reference, it doesn't change the tx just their valuations 
at a point in time for reporting purposes.  From a strict accounting POV 
gnc doesn't need the price db at all because you are allowed to value 
your assets at however much you want.  The tax lady may disagree, of 
course :)

 > > and may be (haven't
> tested) different than a price downloaded via the Finance:Quote module.

It will almost certainly be different, different exchanges have 
different prices, there are people that make a lot of money noticing the 
small differences between exchanges and trading based on those differences.

> And indeed, as the time component is meaningless (yet different in function
> on the price creation method), it shouldn't be stored OR, if for legacy
> reason it should be kept, it could at least be stored consistently (across
> price creation method) using for instance the "neutral time" approach used
> for the post_date.
> If not, any extract of price data (direct SQL, XML, piecash, ...) is
> complex to use.

I don't see the complexity. Why don't you just discard the time 
component as it is essentially meaningless after 24 or 48 hours ?

Isn't your typo a good observation in that it shows that time isn't the 
significant part of the price record ?

P.S. I don't like arguing with you, Sebastien, it is better when we agree :)

-- 
Wm