Best to ditch the word "gender" for inanimates and even many animates, but it's hard because the terminology is deeply embedded in the literature. The endings just help to show what adjective goes with which noun in most cases; in some cases they also show sex, but there's no point in wondering about how a camisa "shirt" is feminine but vestido "dress" is masculine. (Don't like agreement categories in Romance, learn Interlingua).
It's like sorting out why "scissors" is plural (like men's "underwear" is plural for most speakers) but Czech parek is "two hot dogs". Ask for two parky and you get four sausage things.
I reject gender for new adjectives in English and tolerate it for the long-established "blond/e" and "brunet/te", but oddly even publications and style guides that demand gendered "Latino" forms have ruled against the sex distinction for people with brown or blond hair, from what I've observed. (With at least one wanting "brunet" but "blonde" as the remaining acceptable form--go figure.) But if I'm forced to use gender for a few "correct" words, I insist on using them for words and names borrowed from languages I know. (I mean, if we just say "Latino" is an English borrowing like "French" or "immense", then the problem's vanished.)
Some languages have more complex agreement classes. Most Slavic languages distinguish between male/female/neuter nouns (and adjectives that describe them) but also have an animate/inanimate distinction. (And what's in the set of 'animate' varies between languages). They even have female/male/neuter verbs in the past tense (for those with multiple past tenses, just those descended from the l-participle), and Czech at least even has plural past tense verbs that show agreement class, at least in the standard language. They still have (depending on language) profession endings that show sex, but their use is inconsistent and how the absence of sex-marking shows up in terms of "gender" is really a zoo to sort out (if it's even consistent enough to be sorted out).
Bantu languages have far more complex agreement classes, less focused on "male/female" but other often, depending on the language, a bit more abstract. Japanese and Mandarin also has "classifiers".
These are cognitive categories that are formed very early so it's hard to non-native speakers to master them completely. In this sense, they're like TAM systems (tense-aspect-mood), the bane of anybody learning a Slavic language where the aspectual system is very un-Germanic.
= new reply since forum marked as read