| From: | Dmitry Tkach <dmitry(at)openratings(dot)com> | 
|---|---|
| To: | Jean-Christian Imbeault <jc(at)mega-bucks(dot)co(dot)jp> | 
| Subject: | Re: Locking rows | 
| Date: | 2003-02-28 22:25:58 | 
| Message-ID: | 3E5FE1F6.7060504@openratings.com | 
| Views: | Whole Thread | Raw Message | Download mbox | Resend email | 
| Thread: | |
| Lists: | pgsql-general | 
Jean-Christian Imbeault wrote:
> Doug McNaught wrote:
> 
>>If you use a single query to generate the rows, this will not be a
>>problem.
>>
>>If you use multiple queries, execute them all inside a transaction and
>>you will get a consistent view of the database.
>>
>>No locking needed.
> 
> 
> True but what I am worried about is the data changing while I am in a 
> transaction.
> 
> Right now I want to print customer receipts for items that have been 
> ordered. But while I gathering the data, which takes more than one 
> query, a customer might come along and cancel an invoice.
> 
> In that case I would print a receipt for something that was cancelled.
> 
> I need to find a way to avoid this. I thought locking was a way around 
> this ... or maybe I need to change my "business" logic or the way the 
> app gathers the data?
> 
But can't the customer cancel that invoice one millisecond *after* your application is done?
As far as I can see, from the customer's standpoint, it will look exactly the same, won't it?
Dima
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Dmitry Tkach | 2003-02-28 22:28:25 | Re: postgresql caching | 
| Previous Message | Francisco J Reyes | 2003-02-28 22:23:24 | Re: reindex vs. drop index , create index |