Useless mutitasking device check in SDImageCache.m about sdwebimage HOT 12 CLOSED

This is to allow compilation with old SDK (<4.0). This is certainly no longer needed but it doesn't hurt I guess.

from sdwebimage.

This is to allow compilation with old SDK (<4.0). ==> that's exactly what the #if does ;-) Maybe I've not be very explicit. I talked about the if ([device respondsToSelector ..., not the #if. I think it's useless to have the both checks.

This is certainly no longer needed but it doesn't hurt I guess. ==> I guess too. I was reading your class to help me writing my own cache with GCD and Blocks and because I do not like useless code I took 1 min to report it to you ^^

Cheers,

from sdwebimage.

The runtime condition is for when compile for 4.0+ and run it on 3.x

from sdwebimage.

#if __IPHONE_OS_VERSION_MIN_REQUIRED >= __IPHONE_4_0 ==> compile time check, the code will not be executed on device running on 3.x, and can be compiled for those devices => everything is OK.

On device running on 4.0+, the key UIApplicationDidEnterBackgroundNotification will be defined and the notification handler will be called only if the device support multitasking => so the runtime check is useless

PS: excusez moi d'insister c'est juste pour clarifier mes propos et surtout être sûr que je ne me plante pas... On peut continuer cette conversation par mail ([email protected]) si vous voulez ;-)

from sdwebimage.

#if __IPHONE_OS_VERSION_MIN_REQUIRED >= __IPHONE_4_0 ==> compile time check, the code will not be executed on device running on 3.x […]

I think you may be wrong here: if the code is compiled with 4.0+ SDK, it can still be executed on devices running 3.0 depending on deployment version setting. On those devices, the UIDevice's isMultitaskingSupported selector won't be defined and calling it without runtime check will lead to a crash.

from sdwebimage.

Maybe I'm wrong... I wanted to make some tests to confirm that, but the code in the #if is never executed...
I'm probably too tired. I'll give it a try another day and keep you posted ^^

from sdwebimage.

Ok, I'm interested by the result of your test. In the meantime, I keep the code that way. Thanks.

from sdwebimage.

Ok.
In the meantime check out this great blog post => http://mattgemmell.com/2011/12/05/latest-version/
Ok ok, it's a big dilema with apps like yours but you could drive huge amounts of users to update their phone and help all developers to not gone crazy with that kinds of segmentation issues.
;-)

from sdwebimage.

I saw this blog post. I don't totaly agree with it but this is subjective feeling. Anyways, I think it's not the role of a low level library to decide which version of iOS your app will support. Its role is to make sure it will do all the extra-work to ensure it will complet its task in all circonstances. It's why I don't rush on ARC, blocks and GCD for this project as ROI isn't evident.

from sdwebimage.

That's a good point.
I'm agree with you for ARC, but not so much for GCD and blocks because there were introduced with iOS 4.0 and I don't know the exact percentage of users on iOS 3.x but it is really really low, and for me the ROI is evident ;-)

Bye

from sdwebimage.

You mean in term of performance or ease of use? Because in term of perf, I'm not convinced it would change anything actually. If you implement GCD, I would be interested by a comparative bench to see if my feelings were right :)

from sdwebimage.

I mean in ease of use, but a bench could be cool ;-)

I'll finish my library, test it a lot and then push it on github.
Could you make the bench? I didn't have to do it before the next few weeks...

from sdwebimage.