Reply to thread

Message

<blockquote data-quote="Stéphane Lenclud" data-source="post: 1198358" data-attributes="member: 140174">Thanks for the feedback mate.As far as the end result goes and once we make sure the current group is scanned first we don't need any special treatment for the 'all channels' group.However we could end up checking each channel twice. Once through normal groups and once through 'all channels' group.Therefore I would still try to either scan the 'all channels' group or scan the other groups but not both if it can be avoid.The algorithm should go like that:<ul>
<li data-xf-list-type="ul">Scan current group and stop there if it's the 'all channels' group.</li>
<li data-xf-list-type="ul">Scan all groups excluding 'all channels' and current group.</li>
<li data-xf-list-type="ul">Scan 'all channels' group if it's not the current one.</li>
</ul>Then again we could decide performance is not a concern here and we can afford the overhead.That whole channel to number mapping is largely unoptimized to start with anyway.For best performance we ought to have a dictionary of channels indexed by number really.However this is totally outside the scope of those changes.I'll try and see what I can do with 'set current group'.</blockquote>

[QUOTE="Stéphane Lenclud, post: 1198358, member: 140174"] Thanks for the feedback mate. As far as the end result goes and once we make sure the current group is scanned first we don't need any special treatment for the 'all channels' group. However we could end up checking each channel twice. Once through normal groups and once through 'all channels' group. Therefore I would still try to either scan the 'all channels' group or scan the other groups but not both if it can be avoid. The algorithm should go like that: [LIST] [*]Scan current group and stop there if it's the 'all channels' group. [*]Scan all groups excluding 'all channels' and current group. [*]Scan 'all channels' group if it's not the current one. [/LIST] Then again we could decide performance is not a concern here and we can afford the overhead. That whole channel to number mapping is largely unoptimized to start with anyway. For best performance we ought to have a dictionary of channels indexed by number really. However this is totally outside the scope of those changes. I'll try and see what I can do with 'set current group'. [/QUOTE]

Verification