An update from the bug: Seems the behavior is WAI because the execution order of dereference and function calls are not defined. deference of `foo.bar` may happen before or after (which is this case) the function call of `compute_value()`
Spec states following about order of evaluation: > when evaluating the operands <https://go.dev/ref/spec#Operands> of an expression, assignment, or return statement <https://go.dev/ref/spec#Return_statements>, > all function calls, method calls, receive operations <https://go.dev/ref/spec#Receive%20operator>, and binary logical operations <https://go.dev/ref/spec#Logical_operators> are evaluated in lexical left-to-right order. Notably, there is no selector expression, pointer dereference expression (explicit or implicit) and indexing expression (mentioned as example in spec) in that list. On Saturday, December 28, 2024 at 7:01:10 PM UTC-8 Kaiming Yang wrote: > Thanks for the confirmation, I created > https://github.com/golang/go/issues/71054 for this. > > On Saturday, December 28, 2024 at 6:11:52 PM UTC-8 Ian Lance Taylor wrote: > >> On Sat, Dec 28, 2024 at 5:38 PM Kaiming Yang <yax...@gmail.com> wrote: >> > >> > I'm trying to understand execution order of assignment statement, as >> spec mentioned: >> > >> > > The assignment proceeds in two phases. First, the operands of index >> expressions and pointer indirections (including implicit pointer >> indirections in selectors) on the left and the expressions on the right are >> all evaluated in the usual order. Second, the assignments are carried out >> in left-to-right order. >> > >> > I crafted a example about it: https://go.dev/play/p/6RulNQwIk4n, the >> core of that example is this line: >> > >> > phase_2, foo.bar.value = "??????", compute_value() >> > >> > where `foo.bar.value` causes panic due to nil pointer. >> > >> > The execution result shows `compute_value()` is invoked, but `phase_2` >> variable is not updated. Seems like the panic happened after phase 1 is >> complete but before phase 2 started. >> > >> > I assume the panic should happen in one of two phases, but not >> in-between. If panic happened in phase 1, `compute_value()` should not be >> invoked due to it happen later in same phase; If panic happened in phase 2, >> `phase_2` variable should have been changed already. >> > >> > Is the behavior working-as-indented? if so, what am I missing to >> understand it? >> >> Thanks. This looks like a bug to me. I think the panic should occur >> before compute_value is called. When I run your program with gccgo I >> get >> >> original >> panic: runtime error: invalid memory address or nil pointer dereference >> [signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x4037fb] >> >> goroutine 1 [running]: >> main.main >> foo.go:22 >> >> Want to open a bug report at https://go.dev/issue? Thanks. >> >> Ian >> > -- 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 visit https://groups.google.com/d/msgid/golang-nuts/97cb6be8-0c73-4910-9169-274d8ed75945n%40googlegroups.com.
