Josheir : I3 isn't any older than I7, it's just lower in price and perf. Intel has I3/I5/I7 as their "normal" processors pretty much as entry/mid/high power. For what you're doing it hardly makes a difference (it's not like visual studio is power heavy nor are most 2D games). There are very old I7 and very new I3, but in the same age group it always rank I3<I5<I7 with a mix of additional cores and frequency. I'm programming all day long on an I3 on windows 10 and also gaming on it (and not just 2D gaming). You don't need more than that, power has scaled crazy these past few years with the additional cores and not that much in frequency, so the difference between the low and high range in single threaded scenario is not as huge as it once was. If you are curious you can check the cpu benchmark site and compare the relevant processors, the I3 has a cpu mark of 3506 and the i7 of 4324. Also think of I3/5/7 as relative labels, they hold locally but not globally, both of those are portable I3/I7, to give you an idea the best I7 for desktop has a cpu mark score of nearly 15 000, yet it would barely be faster for the mostly single / dual threaded scenarios that seem to be your main use case.
Kryzon : You're mixing both of my points, buying from another country changes the warranty issue but has nothing to do with my statement that it was annecdotal evidence. What i stated was with a sample of 1 computer you can either get 100% or 0% fail rate, it's really irrelevant. Laptops fails, from all brands, ALL BRANDS, and you having one that failed is no reason to discard the brand. I had to return a NEW laptop to one of the biggest shop in paris, their answer was "you were lucky it immediately didn't work, else you'd have to send to the manufacturer for replacement and wait, since it didn't work at unpacking we'll switch it, it happens on nearly 10% of all laptops we sell". That's how commonly a new laptop fails, and while less common the same can be said of older ones, so my suggestions is, when someone asks for what to buy, suggest based on large scope research, not own experience, because with a small sample size it's likely to be harmful and not helpful.