Re^2: why does Perl eval have a strange terminator?

Replies are listed 'Best First'.
Re^3: why does Perl eval have a strange terminator? by haukex (Archbishop) on Jun 14, 2022 at 20:26 UTC
Although I do understand that, it might have seemed more consistent to also make eval { } compound? eval EXPR is often used more like a function (`$sum = eval("3+4");`), and the return value of eval BLOCK is important too (e.g. Bug in eval in pre-5.14). Thus it is consistent with e.g. do BLOCK, map BLOCK, grep BLOCK, and less with `if`, which though it has a "return value", I've very rarely seen it used, and e.g. the return value of `for` should explicitly not be relied upon (I can't find the reference for the latter right now but that's the gist of it). Think of eval like a function with a special syntax for its argument and it makes sense.	[reply] [d/l] [select]
Re^3: why does Perl eval have a strange terminator? by ikegami (Patriarch) on Jun 14, 2022 at 20:55 UTC
I suppose I didn't understand that a simple statement can be, the same time BE a compound statement Something is either a simple statement or a compound statement, not both. But one can include the other. `eval { for (1..2) { f(); } };` is a simple statement. `for (1..2) { f(); }` is a compound statement. `f();` is a simple statement. And `eval` is neither a simple nor a compound statement. As I've already said, `eval` is an operator (like `+`, `and` and `time`), not a type of statement. Operators can include blocks and thus statements: `do` `eval` `map` `grep` `sort` `print` (e.g. `print { ; $fh } $_`) `say` Subs calls to subs with the `&` prototype Indirect method call (e.g. `new { ; "Class" }`) Replacement "expression" of `s///e` Circumfix derefs (e.g. `${ ; \$_ }`) `(?{ })` in regex patterns `(??{ })` in regex patterns	[reply] [d/l] [select]
Re^4: why does Perl eval have a strange terminator? by perlfan (Parson) on Jun 24, 2022 at 20:11 UTC
Try::Tiny requires a `;`, and if you investigate why it is because it is implemented using prototypes. `sub try (&;@) { ... sub catch (&;@) { ... sub finally (&;@) { ...` [download] I am not suggesting "eval BLOCK" is implemented using prototypes, but one could implement their own that would literally become a drop-in replacement (since the `;` is likey already there. FWIW, `(&;@)` means the BLOCK param is coerced into a coderef, then optionally (after that `;`), which allows something like the: `try { } catch { } finally { };` [download] So now the POD seems to make a little more sense.	[reply] [d/l] [select]
Re^3: why does Perl eval have a strange terminator? by LanX (Saint) on Jun 14, 2022 at 21:56 UTC
I think I'm beginning to understand where you're coming from. Try/Catch in other languages is a compound statement, and the new Perl feature is also implemented as such. But `eval {BLOCK};` isn't, it's function like. I think that's at least partly due to sharing its name with `eval "string";` in Perl. Actually it's quite surprising for many coming from other languages, that we call it "eval". Larry often loved to "recycle" concepts. Cheers Rolf _{(addicted to the Perl Programming Language :) Wikisyntax for the Monastery}	[reply] [d/l] [select]
Re^4: why does Perl eval have a strange terminator? by ikegami (Patriarch) on Jun 27, 2022 at 13:29 UTC
I think that's at least partly due to sharing its name with `eval "string";` It's due to it returning a value. That makes it an expression.	[reply] [d/l]