On 05.02.2014 11:01, Axel Dörfler wrote:
On February 5, 2014 at 10:35 AM Stephan Aßmus <superstippi@xxxxxx> wrote:Are you sure you want to set the valid bit when the reply failed?I noticed that is was done inconsistently in the file. In some places, it is set only on success, in others not. The idea could be that when it failed, something is very wrong, and the system should probably not be stressed even more by repeatedly trying to get the value from the server. In any case, I started to make it consistent (change the bit only on success), then reverted and changed it in Origin(), the only place I noticed where it was done so. I can change it either way, but I didn't want to make too many unrelated changes in that commit. What do you think after I explained one possible rational?Now I'm confused :-)
I doubt that. :-) I just gave you one reason to include in the consideration, I am not even sure myself whether it's a good reason. The API client will never even see any failure, so much is for sure. The question is only, whether the retrieval over the link should be retried or not, the next time the value is requested.
The only scenario where having a flaky link is likely would be when it's run over the network. In that case, setting the flag unconditionally may result in an inconsistent state between view and server. The question is if that really matters as much; the drawing should be messed up either way, right?
Yep. Well, it depends... Best regards, -Stephan