| MMMMMMMMMMMM 111222333444 | MMMMMMMMMMMM 111222333444 | MMMMMMMMMMMM 111222333444 | MMMMMMMMMMMM 111222333444 | | (this could be used in a multihead setup)
Personal opinions ahead, but they might be of interest as real-world experience with dualhead setup :-) In my experience, dual-head doesn't combine flawlessly with the "zoomed & stack" approach because of human issues. Firstly, (I'd imagine that like me) most people with multihead setups have them to display more things, rather than the same numbers of things bigger. Certainly I tend to find myself using two/three "full length" columns and maybe just one "slave" column quite a lot. Secondly, if you put two screens with the join directly in front of you you really want to be able to put the stuff you're primarily working on near the centre rather than always off on the far left. Thirdly, even if you have one "zoomed-ish" client it's still useful to be able to rearrange the order of "slave" clients, eg, so you can put two things you want to compare by flicking your gaze between them close together. Lastly, again with this extra space you might want to use a mix of "naturally tall clients" (eg, text windows) and "naturally wide clients" (eg, time series graphs) on one view. I've tried to partly address these issues in my large numbers of clients patch (such as in the still flakey "move client to position x in displayed clients" functionality), but I'm not particularly satisfied with the result; the patch is a hack to make things more usable for me without completely rethinking the layout, its code and implications from scratch. That'd be very interesting to do, but I myself don't have time at moment. Note that this is just pointing out that with multihead the "human issues" are at least as important as extending the single-head paradigm in a logical and consistent way. cheers, dave tweed
