So, how often do you depend on this particular scoping behavior, in Go or JS? Do you have any example of where you intentionally rely on the current behavior, that might break with the new one?
I think it's important to emphasize that we do not know of a *single case* where the current behavior was usefully taken advantage of. Even strong opponents of the change failed to produce any - they just created synthetic (and extremely obfuscated) cases that would change behavior, without any reference to real code these cases where taken from. You keep asserting that you have to remember different sets of rules - I just don't think that's true, in any real sense. I think you'll find that in reality, it will have literally no effect on you - except that it is not unlikely to fix a bug or two in your existing code base. Because it turns out that no one actually thinks at that level about scoping. On Sat, Sep 30, 2023 at 6:35 PM Victor Giordano <vitucho3...@gmail.com> wrote: > Thanks Alex, your insight was very helpful. > > Allow me to share the feeling I have => I still struggle a little in my > mind... I craft web fronts in javascript, and back in golang (and scala or > java). With this change I will have two different scoping rules... and I > feel I don't need it (because I learn how to use it well) nor ask for > them... but it is okay.. changes work that way. The bottom line is: I was > already comfortable with the language scope rules. > > El sáb, 30 sept 2023 a las 12:37, Axel Wagner (< > axel.wagner...@googlemail.com>) escribió: > >> This has come up during the discussion already. I don't know enough about >> other languages to speak with confidence, but apparently there already is >> precedent for this and/or some languages are at least considering making a >> similar change. >> >> Note that scoping rules already vary between languages - in some cases, >> pretty widely. For example, Python is not block scoped - this works: >> def f(): >> if True: >> x = 42 >> print(x) >> And Perl has dynamic scoping, where variables can be scoped to a call >> stack, instead of any lexical element of the source code. Javascript, as >> far as I know, is pretty notorious for having extremely idiosyncratic >> scoping rules (e.g. it is a common confusion how `this` is scoped in >> different contexts) and also has several different kinds of declarations >> with AFAIK slightly different scoping rules. >> In C, a symbol is only scoped to a single file (as far as the processor >> is concerned) and in there, only from its declaration onwards, while in Go, >> a package-scoped definition can appear in any file of any package. And >> speaking of C, it doesn't have closures at all, so the scoping rules of >> loop variables are *already* very different. >> >> So I think the case that you *currently* don't have to be aware of how a >> language is using scoping is pretty vastly overstated. What's more, the >> majority of programs won't actually be affected by this - and for those >> that are, it seems that empirically, the new rules will align more closely >> with what people expect subconsciously. >> >> I don't think you should worry too much. Run your tests with the new loop >> variable semantics in Go 1.21 to see if everything still works. Most >> likely, it will - or you will discover a bug in your current code. >> >> On Sat, Sep 30, 2023 at 4:58 PM Victor Giordano <vitucho3...@gmail.com> >> wrote: >> >>> Hi gophers! How you doing? >>> >>> In my work we recently discuss with a buddy about this article >>> <https://go.dev/blog/loopvar-preview>. I need to talk about this.... >>> >>> I appreaciate that golang makes a lot of effort on this. Working with >>> clousures cames with perils and the go vet tool emiting a warning is a >>> great thing. >>> >>> Althought I do not think that chaning language semantics is something >>> 100% good, see my point: As long I use several languages for making a >>> system this "scope rule" will mismatch with other languajes, for example, >>> javascript. >>> >>> So from now onwards I shall be aware that for loops have variables >>> scopes in one fashion in Golang and in other fashion in another >>> languages... so at the end of the day, I feel like is adding some burden. >>> If for development we can use a single and sole language (Sauron don't read >>> this :P !) I would think this change is good, but that is not the case in >>> now-a-days. I try to think that with time, perhaps other languages evolves >>> it this way... but.. ¿what if I wanna have my "for loop" with legacy scope >>> (block scope, not per iteration scope)? >>> >>> So... I expect to the readers to talk with me, showing what they see and >>> feel. Is not my intention to generate hard feelings at all. I'm a person >>> that work crafting system in group with others.. I have had terrible >>> nightmares working with Scala where implicits and invoking methods as they >>> were feilds were terrible ideas for working in group (perhaps not if you >>> work solo). And this feature recalls me those feelings. >>> >>> Greetings >>> Víctor. >>> >>> -- >>> You received this message because you are subscribed to the Google >>> Groups "golang-nuts" group. >>> To unsubscribe from this group and stop receiving emails from it, send >>> an email to golang-nuts+unsubscr...@googlegroups.com. >>> To view this discussion on the web visit >>> https://groups.google.com/d/msgid/golang-nuts/5f31be67-d246-4778-a373-69d525772974n%40googlegroups.com >>> <https://groups.google.com/d/msgid/golang-nuts/5f31be67-d246-4778-a373-69d525772974n%40googlegroups.com?utm_medium=email&utm_source=footer> >>> . >>> >> > > -- > V > -- You received this message because you are subscribed to the Google Groups "golang-nuts" group. To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/golang-nuts/CAEkBMfGuOXw9nBnMDadG%2BwiY90VvzZLdX5Am1Y%2ByUPM_uTR3-g%40mail.gmail.com.
