On Tue, Jul 12, 2016 at 2:40 PM, Denis Magda <[email protected]> wrote:
> Personally, I don’t get why we need to have specific timeout related > exceptions. My preferences is to stop this practice starting from > IgniteDataStreamer and create generic TimeoutException that can be used by > the new features later. > Disagree. We should either fix it everywhere, or remain consistent everywhere. From the API standpoint, it will be very confusing if we don’t. One way to fix it everywhere, and to preserve backward compatibility, we could have all specific TimeoutExceptions inherit from the generic TimeoutException and then deprecate them. > > — > Denis > > > On Jul 12, 2016, at 2:35 PM, Dmitriy Setrakyan <[email protected]> > wrote: > > > > If other Timeout Exceptions are specific, shouldn’t this one be specific > to > > data streamer as well, for consistency reasons, e.g. > > IgniteDataStreamerTimeoutExcepiton? > > > > On Tue, Jul 12, 2016 at 2:20 PM, Vladislav Pyatkov < > [email protected]> > > wrote: > > > >> Dmitriy, > >> > >> I discussed that with Denis and we have decided create new Exception > type > >> by some reasons: > >> 1) The exception need to be unchecked (the behavior means serious > problem > >> in grid) hence, TimeoutException will implement IgniteException (as > >> RintimeException). > >> > >> 2) Other timeout exception, which follow the logic, are specific (for > >> example TransactionTimeoutException). > >> > >> So TimeoutException will be RuntimeException, will not be specific for > >> DataStreamer and can be reuse. > >> > >> On Tue, Jul 12, 2016 at 1:54 PM, Dmitriy Setrakyan < > [email protected]> > >> wrote: > >> > >>> Thanks Vlad. > >>> > >>> At a high level, the changes look OK. However, I am not sure about > >>> TimeoutException. Don’t we already have other timeout exceptions in > >> Ignite? > >>> Can we reuse them? > >>> > >>> D. > >>> > >>> On Tue, Jul 12, 2016 at 1:51 PM, Vladislav Pyatkov < > >> [email protected]> > >>> wrote: > >>> > >>>> Dmitriy, > >>>> > >>>> I have added the description of changes in the JIRA ticket. > >>>> > >>>> On Tue, Jul 12, 2016 at 1:28 PM, Dmitriy Setrakyan < > >>> [email protected]> > >>>> wrote: > >>>> > >>>>> Hi Vlad, > >>>>> > >>>>> Can you please list the API changes in the ticket, so others can > >> review > >>>>> without digging in code? > >>>>> > >>>>> Thanks, > >>>>> D. > >>>>> > >>>>> On Mon, Jul 11, 2016 at 1:14 PM, Vladislav Pyatkov < > >>>> [email protected]> > >>>>> wrote: > >>>>> > >>>>>> Igniters, > >>>>>> > >>>>>> I have implemented timeout in DataStreamer by issue IGNITE-3055 > >>>>>> <https://issues.apache.org/jira/browse/IGNITE-3055 > >>>>> (*IgniteDataStreamer > >>>>>> can't be timed out*). > >>>>>> > >>>>>> That lead the change public API: > >>>>>> > >>>>>> 1) Added set and get methods (IgniteDataStreamer.timeout(long) and > >>>>>> IgniteDataStreamer.timeout()) for establish. Default value is -1, > >>> means > >>>>>> unlimited time. > >>>>>> > >>>>>> 2) Added new TimeoutException (inherited IgniteException). The > >>>> exception > >>>>>> will thrown when timeout will be reached (may be take place in > >>>>>> IgniteDataStreamer.addData, IgniteDataStreamer.close and > >>>>>> IgniteDataStreamer.flash operations). > >>>>>> > >>>>>> Dmitry, colleagues, please comment or approve. > >>>>>> > >>>>>> > >>>>> > >>>> > >>> > >> > >
