reminder2

Susanne Lefvert lefvert at mcs.anl.gov
Thu Jul 3 13:55:41 CDT 2003


Bob wants us to do try except on soap calls in AccessGrid.VenueClient and 
the re-raise our own defined errors instead of catching all exceptions.  
This is what we said on the mud:

Bob [to you]: where's the call you're talking about?
lindsay idles: meeting
>bob, AccessGrid/VenueClient/GetPersonalData
** First event: 1:24pm 07/03 Last event: 1:24pm 07/03
** Bob spoke directly to you.
You [to Bob]: AccessGrid/VenueClient/GetPersonalData
Bob says, "so I don't see those calls in a try block"
>bob, and that method is called in VenueClientUIClasses.OnExpand
You [to Bob]: and that method is called in VenueClientUIClasses.OnExpand
Bob says, "you should probably try that first call, verify you got a url 
and
not an exception, then make the second call"
>bob, I will never get a url
You [to Bob]: I will never get a url
>bob, because GetDataStoreInformation does not exist on the old client
You [to Bob]: because GetDataStoreInformation does not exist on the old 
client
Bob says, "right, so the first call will fail and you'll explicitly catch 
it at
that site"
>"ok
You say, "ok"
Bob thinks it's best to do the checking for bad remote calls at the site 
of the
call, rather than further up
Dave hrm
>bob, if it always throws an exception as it should one try block should 
be enough
You [to Bob]: if it always throws an exception as it should one try block
should be enough
>bob, else I would just have to raise it again to get the ui to react
You [to Bob]: else I would just have to raise it again to get the ui to 
react
Bob says, "right, but semantically it makes more sense to me to have the
getpersonaldata call return None if datais not availalbe, or [] if data is
available but none currently exists"
Bob says, "rather than catch some transport-dependent exception"
Bob note that with teh catchall exception you could be getting exceptions 
for
other cases too, like typos in the called code
Ryan has disconnected.
>bob, right, that is probably more precise but I didn't know what 
exceptions I could expect so I just catched all to be safe
You [to Bob]: right, that is probably more precise but I didn't know what
exceptions I could expect so I just catched all to be safe
<IDLE>
Bob [to you]: that's why I argue for catching them down at the remote 
call, and
either not using excpetions to pass status back or to pass explicitly 
defined
exceptions
>:likes explicitly defined exceptions better
zuz likes explicitly defined exceptions better
Bob too





More information about the ag-dev mailing list