Weird list

[scratchmodification]

I just tried a custom block and it is supposed to report a list containing

• if the command failed.
• the error (only appears if the handler was run).
• data that the command or handler reported.

This is the definition:

---

Apparently, works fine:

until

.

And that's not all:

,
, and it continues.

What I could see is that sometimes when I use a reporter or predicate with that list, such as , it reports the same list! My mind's going to explode rn.

---

Think it's a JS bug, but I'm not sure. Hope this gets fixed bc I'm making a category for the Better saga.

[bh]

I'm having a little trouble reading your mind (an important skill for teachers), but your use of JOIN is incoherent, so I think you're confused about how the CALL surrounding it works.

Okay, here's my guess: You meant to say

but COMMAND doesn't take any inputs, so even that wouldn't mean anything.

No matter what you try to wrap around it, COMMAND is a command; it doesn't report a value, and so you can't use that nonexistent value in a larger expression.

So, your basic idea of

is good, but you can't run COMMAND inside the REPORT. You have to run the command, and if you actually reach the REPORT, then the command worked:

Alas, even this isn't quite good enough. Read the definition in the library:

You're not doing that ERR_RESET. I don't know what that does, exactly, not having read the relevant code in the Snap! source files, but it wouldn't be there if it weren't important! So you have to do this:

Okay, you take it from there!

[scratchmodification]

@bh

Okay, here's my guess: You meant to say

but COMMAND doesn't take any inputs, so even that wouldn't mean anything.

Nope, error upvar is set to "error" when the command is run. I want to set it to "" so that's why I use
. So, for example, I want to convert
to , to finally call it as .

@bh

is good, but you can't run COMMAND inside the REPORT. You have to run the command, and if you actually reach the REPORT, then the command worked:

And what if the command reports something? I'm going to lose that value.

@bh

You're not doing that ERR_RESET.

I tried doing err_reset but doesn't work either.

---

@scratchmodification

Apparently, works fine:

until

As can you see, the real problem is the list. I need another solution.

---

Thank you for the advice though

[bh]

@scratchmodification

And what if the command reports something?

By definition the command doesn't return anything! That's what makes it a command.

There is a separate block

(note its round shape) for calling reporters and catching errors.

@scratchmodification

As can you see, the real problem is the list.

The real problem is that you're putting a command block in a CALL. That's known to give strange results. If it's a command block, you have to RUN it. If you want to CALL it, it has to be a reporter (or a script with a REPORT block).

P.S. Would Snap! be a better language if it only had reporters, like Scheme? Maybe it would, maybe it wouldn't. But it is what it is.

[scratchmodification]

@bh

By definition the command doesn't return anything! That's what makes it a command.

So, what is this block??: . You can put it in commands.

@bh

There is a separate block

(note its round shape) for calling reporters and catching errors.

I've tried this block. Doesn't solve the problem.

@bh

That's known to give strange results.

Really?

[helicoptur]

This happens sometimes when you have the report block in the definition of a “safely try” block. It’s quite strange, but basically the report block overrides all its parent scripts so that it can report the value. I’ve encountered this back in the days when JS wasn’t disabled.
Try using the reporter version of the gray PRIMITIVE block instead.

[scratchmodification]

@scratchmodification

@bh

There is a separate block

(note its round shape) for calling reporters and catching errors.

I've tried this block. Doesn't solve the problem.

I said it doesn't work either.

[bh]

@scratchmodification

So, what is this block??: . You can put it in commands.

Using REPORT in a command block is an error. Apparently we don't give an error message for it, but we should.

@scratchmodification

I've tried this block. Doesn't solve the problem.

Perhaps you're using it wrong, e.g., not calling err_reset? The reporter SAFELY TRY in the library doesn't use it, to my surprise; it has a command SAFELY TRY inside.

I guess I don't really understand why you want to rewrite this library...

[scratchmodification]

@bh

Perhaps you're using it wrong, e.g., not calling err_reset? The reporter SAFELY TRY in the library doesn't use it, to my surprise; it has a command SAFELY TRY inside.

I mean if I use that reporter doesn't solve the problem (the list is still weird).

@bh

I guess I don't really understand why you want to rewrite this library...

It's like the safely try block but it returns a list containing information:

@scratchmodification

I just tried a custom block and it is supposed to report a list containing

• if the command failed.
• the error (only appears if the handler was run).
• data that the command or handler reported.

[warped_wart_wars]

@bh

P.S. Would Snap! be a better language if it only had reporters, like Scheme?

Who would know without trying it?

[scratchmodification]

Disagree. I like using commands to make definitions.

[warped_wart_wars]

In Scheme, expressions can act like reporters:

(if (< x y)
    (expt 2 x)
    (expt x 2))

or commands:

(if (< x y)
    (print "x is less than y")
    (print "x is greater than or equal to y"))

[scratchmodification]

You can use

[scratchblocks]
(call ({­::control}::grey ring)::control)
[/scratchblocks] reporter.

[warped_wart_wars]

???

What do you mean?

[scratchmodification]

I can't use images bc Snap! website is down.

[warped_wart_wars]

Try using

@warped_wart_wars

Snap! 6.9.2 Build Your Own Blocks or Snap! Build Your Own Blocks.

[warped_wart_wars]

Do you mean that this works?:

[scratchmodification]

Yes

[warped_wart_wars]

That's because the

...
report (... :: grey) :: control cap

is a reporter disguised as a stack of commands

[scratchmodification]

Ok, so do you have a solution?