In application development I would normally do this type of thing inside
CONFIG::dev
blocks, or similar, so having the possibility, as described, to exclude
development-only code (extra type checking, null checks etc) would
definitely also be helpful PAYG-wise from within the framework.
The general issue here I think is that using this approach for the
framework requires a dev build of the framework and a separate
produciton-ready build, so the approach of using GCC to avoid this via
annotations sounds like a great solution to that. I don't know how we could
do this on the swf side though without having two framework builds?
btw React/React-Native has something like this as well iirc, including
removal of a bunch of runtime type-checking support in the release builds,
so having something like this seems like it would be a good thing to for
comparison with others.
I terms of the 'type check' in this case, personally I think it could be as
simple as
ba = bytes ? bytes as ArrayBuffer : new ArrayBuffer(0);
if (bytes && !ba) throw new Error('unexpected constructor argument type');
I think that could be used for both swf and js.
On Wed, Jul 12, 2017 at 8:56 AM, Alex Harui <aha...@adobe.com.invalid>
wrote:
> I've mentioned the notion of debug-mode beads on the mailing list a couple
> of times. It would be great to see this idea explored more.
>
> By definition, now that you've debugged your code, any type-checking code
> path becomes unused in production, which is what we don't want. But
> adding code paths that go away in production, or swapping in different
> beads during development that have more code paths that check for common
> mistakes is certainly within the charter. We want to maximize developer
> productivity.
>
> I think Google Closure Compiler supports a debug flag you can use in
> COMPILE::JS blocks that gets dropped in production. If we need to mimic a
> similar flag for SWF so code can compile outside of COMPILE::JS blocks
> that might be ok.
>
> -Alex
>
> On 7/11/17, 1:07 PM, "Harbs" <harbs.li...@gmail.com> wrote:
>
> >I just wasted over an hour because I was initializing a BinaryData with a
> >string instead of an ArrayBuffer.
> >
> >I would like to add a typecheck to the bytes argument in the BinaryData
> >constructor to throw an error if something other than an ArrayBuffer is
> >provided. We cannot use strict typing to catch this in the compiler,
> >because the argument is different for SWF and JS. Is this a violation of
> >PAYG? It’s sort-of just in case code, but not really because it’s
> >protecting against errors.
> >
> >Thoughts? Other solutions?
> >Harbs
>
>
