In the context we're discussing, we're explicitly taking a job to sample many things.
Given the recent specialization in programming (web backend v. frontend, "mobile", devops, etc.), which is a continuing trend from previous decades, this seems sensible. I advocate for a T shape in skills, but before you pick what you go deep in, get somewhat established in the wide. Fresh out of school you just can't have the width yet to know what you want to and should go deep in yet, not even if you've had a job the whole time.
Actually, I suppose I ought to raise that up as my true point: Go a bit wide before you select what to go deep in. Try not to just float into something without thought. Width should emphasize learning things that have demonstrated staying power, but you want to go deep on something that you expect has future power. It probably shouldn't be the current hottest fad tech, but the current hottest fad tech is probably at least in the area of a good future choice; it's more interesting why the current fad is a fad than the specific fad. If you do go for something that became super hot in the last 6 months to a year, try to figure out how to leverage that into learning about fundamentals even so. There's ways at both big companies and startups to succeed and fail by this metric.
Of course, on the flip side, you don't have to get it perfectly correct. You can't go deep in everything all the time, but "pi-shaped" Π is feasible in addition to T-shaped too. (That's capital pi, for those who may not have seen it before.)