sl75
Posts: 2,092 
Likes: 1,245
|
Post by sl75 on Dec 10, 2014 17:47:09 GMT
It's weird because your 'sold parts' report (sell page, sold parts iframe) never (afaict) shows a double sale, but the transaction report manages to .. since the transaction report doesn't show 'who it was sold to', I can't check and see if the double sales were in fact to the same person/bot. I suspect you might be right - the SM (buy parts) page is one clunking antique which hasn't worked right for ages (shows parts which don't match filters, doesn't show parts which do, etc. etc) so it would not be hard to buy more than once, especially if a bot was doing it at high speed (there is a random number injected into the page requests IIRC, presumably to try to prevent caching problems, but it is onbly a 4 digit random number so it's not going to be awfully reliable). It should not be possible .. a buy transaction ought to be an indivisible unit, and the transaction report ought line up with the sold parts report (allowing as how the latter drops all parts from loans which are repaid, or defaulted. 8<.) But then Fairly Colourful are masters of the impossible .. On occasion I've attempted to click on the "buy" buttons of a more than one part of a loan for sale in relatively quick succession. The overlapping requests often seem to result in a "this loan part has already been sold" for some or all of the requests. Not investigated it in detail (and I doubt as an end user I would have tools to do so), but it seems to suggest some data leakage between multiple simultaneous (or overlapping) purchase requests - maybe part of the back-end code is incorrectly assuming that one user will only be attempting to buy one loan part at a time? ... or using some session variables for things that are specific to a single request?
|
|
mikeb
Posts: 1,072
Likes: 472
|
Post by mikeb on Dec 10, 2014 19:33:18 GMT
This double sale of loan parts happened to me recently, but it was only today when the second shoe dropped  I apparently sold loan part 5843092 on 1st December, but it didn't show up in "sold loan parts". Had no idea it had gone, but did have a spurious ~£20. I sold it again on 2nd December, to "Ad*m", and this time it did appear on "sold loan parts". This is when I plucked the transactions for that loan part from my statements ... Both sales appear on my statement, and for the last few days I've had a spurious ~£20 floating about. Today I was pushed out of 6 auctions, and was only able to push back in on 5, because the sneaky reconciliator has nicked ~£20 to settle up. The 1st December sale has now gone from my statement, as if it never happened, and everything balances again. |
|
|
|
Post by GSV3MIaC on Dec 10, 2014 20:44:05 GMT
Ah, so what we actually had here was a 'sale' which only part happened (on 1st Dec) - you got the money, but apparently nobody got the part (one wonders where the cash came from?!). Seems Fraction Computing have managed to boldly split what no man has split before. Previously I think we have heard from people who bought parts, and then had them automagically re-sold (so presumably hadn't really bought them for whatever reason). Ooops ^ 2
|
|
blender
Member of DD Central
Posts: 5,719
Likes: 4,272
|
Post by blender on Dec 11, 2014 0:09:55 GMT
I think that isolated incident of the re-selling without asking is not connected.
We have not heard of people buying a part, paying the money, but not getting the part. I know that forum people tend to be SM sellers rather than buyers, but I think we would have heard of it.
What seems to happen is that someone tries to buy the part but does not get it and their account is untouched. But that request somehow results in the seller getting paid for the part without the part being transferred - so that available funds is credited but amount lent is not debited. The debit would fall on some reconciliation account belonging to poor old Funds Compromised. But that is not a double sale, and it seems to relate only to parts sold and paid for twice but the part transfer and the corresponding debit happening once. Experience suggests that the error sequence is related to two purchase attempts which involve processes which overlap in time, whether by the same or different purchasers, or in my first case the seller attempting to retrieve some parts from sale while someone is trying to buy them. (Mikeb's experience is a puzzle - but I wonder if the two purchases were coincident and the reporting of the second was the end of the error sequence. Perhaps the system uses advanced quantum-based processing in which a sale is not decided and real until it is observed).
If it happened with a single purchase attempt then there would be much more of it and it would be much easier to find - or so it seems to me.
But I only press mouse buttons on the webpage - and whether bots have a greater ability to Fundamentally Confuse the machine is not known to me.
|
|
