[Jira-196] Support for postfix and prefix increment and decrement operators. about chapel HOT 2 OPEN

Historically, we shied away from supporting these operators in Chapel's original design because we were generally trying to avoid supporting operators that modified things at the expression level (primarily to reduce the number of cases in which we'd need to understand read vs. write semantics on expressions in the compiler). I.e., the only way to modify things was through assignment operators and function calls. (We also avoided expression-level assignment for this reason, though the rationale for that decision is arguably stronger, since it helps avoid common = vs. == mistakes and could result in confusion in contexts like named argument passing). I also was never particularly fond of (or good at) distinguishing between ++a and a++.

All that said, my personal negativity toward them isn't as strong today as it was then. In the past I've argued "you can always define methods that do this so there's no need for language support" and while that's true, it also suggests that supporting them wouldn't be all that bad / different from doing so via methods, while being succinct and familiar to C-family programmers. All that said, I'm also am not eager to add them without a good/clean story for how they'd be defined by a user on their own type

In the perf meeting today, we (@ronawho, @ian-bertolacci, @benharsh) discussed a few use cases that might motivate supporting them:

• If we move toward supporting operator overloads for common atomic operations, then += would correspond to a .add(), but if we had ++, that could support the common case of a .fetchAdd(1) nicely.

• @benharsh points out that these could be useful in bucketing patterns like ISx as well.

from chapel.

This SO Q&A which seems to summarize Swift's choice not to support these operators was interesting to me.

from chapel.