Re: AddGadgetItem image bug, all OS
Posted: Sat Feb 11, 2017 12:57 am
Lets add a bit of perspective:
With your example code from above I did a few measurements which compare your use case (a large number of different images) vs. the much more common use case (same number of items, but with only a few distinct images) with our caching behavior. I measured how long it takes to populate the gadget with the items.
Windows 7:
I also looked up the history of this particular functionality: It has been virtually unchanged since I ported the library from Assembly back in 2006, and as far as I can remember there have not been any complaints about it so far. So your statement that "everybody" is affected by this must be taken with a grain of salt as well.
Please note that I am not saying that your use-case is not valid. You just have to see the other side of it too. We have to make such balancing decision within the PB commandset all the time and its impossible to always satisfy everybody. If you have needs that are outside of the norm, you may have to resort to using an API solution. This way you have all the freedom to go the route you want to without being limited by our design. And its not that difficult either, because as you have demonstrated above the API way can be mixed with regular PB commands quite easily (which is also something we sometimes go to quite some length to ensure).
I understand your frustration but this is not just about you. Sorry.
With your example code from above I did a few measurements which compare your use case (a large number of different images) vs. the much more common use case (same number of items, but with only a few distinct images) with our caching behavior. I measured how long it takes to populate the gadget with the items.
Windows 7:
- Many images: 6 seconds
- Few images: 3 seconds
- Many images: 17 seconds! (there you have the reason why we added caching in the first place)
- Few images: 3 seconds
I also looked up the history of this particular functionality: It has been virtually unchanged since I ported the library from Assembly back in 2006, and as far as I can remember there have not been any complaints about it so far. So your statement that "everybody" is affected by this must be taken with a grain of salt as well.
Please note that I am not saying that your use-case is not valid. You just have to see the other side of it too. We have to make such balancing decision within the PB commandset all the time and its impossible to always satisfy everybody. If you have needs that are outside of the norm, you may have to resort to using an API solution. This way you have all the freedom to go the route you want to without being limited by our design. And its not that difficult either, because as you have demonstrated above the API way can be mixed with regular PB commands quite easily (which is also something we sometimes go to quite some length to ensure).
I understand your frustration but this is not just about you. Sorry.