NYCPHP Meetup

[Cliff Hirsch]

> Seems like the real problem here is that the application itself is not
> aware of which requests have been submitted already and which haven't.
> Of course you could do some things to skip out the error (error
> supression with @, using "insert IGNORE into ..." to prevent errors on
> duplicate keys, etc.), but I think you will make your life easier by
> addressing the root of the problem: if the page is reloaded, make sure
> the code isn't run a second time, unless that's what's supposed to happen.
> 
> In my systems, I'll assign a unique ID (based on microtime) to each
> submitted request (added as a hidden input in forms, else appended to
> the query string), to be stored in the database.  For each request, the
> system checks to see if the ID has been submitted; if so, the action
> code is skipped and only the display code is run; if not, the request ID
> is recorded in the databse, and the action code is processed as normal.
>   Garbage collection routines keep the request-ID cache from getting too
> big but keep the contents long enough to avoid false negatives.
> 
> That's just one way to do it.  Others here can probably recommend better
> methods, but I think you have to find some way to do it.  You probably
> don't want to go down the path of just supressing errors all over the
> place, because when a real error surfaces you need to be able to see it
> and handle it properly.
> 
> - Allen

Thanks. Your comment made me realize I already do this, I just had it
disabled for testing -- late day brain fart! But I tend to be paranoid, and
check at probably too many levels.