If buy() fails, can we get mall_price() updated?

[dj_d]

This is a scenario from eatdrink.ash. You use mall_price to figure out how much you should pay. You buy the item for up to that amount, actually a bit more:
buy(1, it, mall_price(it)*1.25)
Buy() fails because the price has gone way up. The problem is, mall_price still mistakenly thinks the price is low. I want to recalculate and make sure it's a good value at the new price, then try again.

So It would be great if mall_price updated the price on an item it if a buy() command for that item fails because the item was unavailable for that price, and the limit price was equal to or greater than the mall_price amount. Or at the very least, if buy(quant, it, price) fails, set mallprice(it) = max(mallprice(it), price+1).

Did that all make sense? Any reasons that won't work? The workaround in eatdrink.ash otherwise is pretty ugly.

 

[Bale]

Asked about that previously. Jason pointed out that would allow unlimited mall price searches per day by simply sending buy() requests with a low price limit that you know will fail.

What I do if my purchase fails is to raise the price limit to the most that makes sense and only try to purchase 1 of the item. Then if it fails, I remove it from consideration altogether. I'm not happy about it, but... meh.

 

[dj_d]

No, the situation I described doesn't do that, since it only forces an update if the buy request is at or above mall_price. In plain english, if you make a good-faith attempt to buy at the ostensible mall price and fail, it's reasonable to update you with the current mall price.

 

[Bale]

I understand now. That sounds reasonable. If a buy request is made for a price that is at or above mall_price(), then it should update on failure...

I hope that Jason and Veracity agree with us.

 

[dj_d]

The alternative is to wrap mall_price() and, in the case where a buy() fails, return the amount of the attempted buy + 1. But that's more likely to generate a second purchase failure than just finding the correct new mall_price().

(this isn't theoretical - eatdrink just hung on twinkly wads for this reason yesterday; there were some of them super-cheap, with a limit of 9 or so).

 

[dj_d]

Argh. I'm writing a workaround for eatdrink, but it's ugly.

 

[jasonharper]

Implemented in r7590.

 

[dj_d]

Yay! I just knew that as soon as I wrote the workaround, it would become unnecessary. Thank you, universe*, and thank you, JH**.

* sarcastically
** seriously

 

[Bale]

I guess this means I should remove my workaround from Universal_recovery also...