Flush cached join results vs Daniel Wood's cartesian refresh
So I'm trying to adapt/learn from starter solution scripts and I'm pretty happy with the results so far. One concern of mine however is the frequency of "Refresh Window [flush cached join results]". I've read on several blogs/articles that this can be a performance killer on a WAN. In this blog post (and subsequently on many other blogs/posts) Daniel Wood outlines a method of using cartesian joins (a method which, seemingly, gets an improvement in the comments (see Jason DeLooze's comment) to only refresh the relationship in question.
Now I'm still very much trying to get a handle on the basics of what needs refreshing and why, but considering that I'm currently adapting filemakers scripts, it would be great if I could get this issue right as I implement, rather than having to go back and substitute all those refresh triggers with this Mr. Wood's method. That blog post, despite the bugged dates on the comments, appears to be over 3 years old (FMP 11). I've found many posts on fmforums and elsewhere in the meantime regarding this issue, though most remain unanswered/inconclusive.
Is anyone out there familiar with this issue? Is this cartesian join method still used in fmp solutions? Or are there any known issues with it? I'm particularly inclined to think refresh is still an issue over WAN's since Filemaker's solution only run the refresh step on desktop devices and not ios/web (but that's only a guess- maybe the only use it on desktops for other reasons).
Many thanks in advance.