traces not being stripped out

[Harbs <ha...@gmail.com>]

I just noticed that traces are currently outputting in minified release code.

I also noticed code related to Vectors which should not be there. I suspect that there’s more code that’s not being stripped as it should be.

Has anyone recently looked at this? It definitely used to be stripped out. So something must have changed.

Harbs

Re: traces not being stripped out

[Harbs <ha...@gmail.com>]

Done.

https://github.com/apache/royale-asjs/issues/1163

> On Nov 8, 2021, at 6:29 AM, Greg Dove <gr...@gmail.com> wrote:
> 
> For the 'traces not being stripped out' issue, I did not look into that. If
> you have a minimal example please open an issue and assign it to me, I will
> look into it next weekend.

Re: traces not being stripped out

[Josh Tynjala <jo...@bowlerhat.dev>]

Thanks for figuring that out, Greg!

--
Josh Tynjala
Bowler Hat LLC <https://bowlerhat.dev>


On Sun, Nov 7, 2021 at 8:29 PM Greg Dove <gr...@gmail.com> wrote:

> With the latest compiler updates, @royalesuppressexport should be working
> again - I can now see the Vector stuff being dead-code-eliminated again if
> it is not used, and 'org.apache.royale.utils.Language' and its members are
> also no longer exported.
> I did add some sdk level tests to monitor this feature, in the reflection
> lib, so if anything changes and it no longer works, we should get to know
> about it.
> It was quite tricky to get this to work without changing the way it is
> supported via ASDoc comment, because I think we are using those doc-comment
> directives in ways that were perhaps never envisaged at the outset, but I
> did find a sneaky way to do this that seems to work. It is using metadata
> 'under the hood' so it persists in the swf definitions inside a js swc,
> which was the main problem that arose with the 'dynamic' nature of the
> export instructions for Closure.
>
> For the 'traces not being stripped out' issue, I did not look into that. If
> you have a minimal example please open an issue and assign it to me, I will
> look into it next weekend.
>
>
> On Mon, Nov 1, 2021 at 2:28 AM Greg Dove <gr...@gmail.com> wrote:
>
> > For the traces issue, I am not sure, but for Vector and others I believe
> > that was related to compiler changes which meant that
> @royalesuppressexport
> > no longer works, because the way things are being 'exported' changed.
> >
> > I will try to look into that as soon as I can. Josh and I had discussed
> > this briefly, so this feature likely needs to use Metadata now iirc. I'll
> > create an issue later today and assign to myself.
> >
> >
> >
> > On Mon, 1 Nov 2021, 12:46 am Harbs, <ha...@gmail.com> wrote:
> >
> >> I just noticed that traces are currently outputting in minified release
> >> code.
> >>
> >> I also noticed code related to Vectors which should not be there. I
> >> suspect that there’s more code that’s not being stripped as it should
> be.
> >>
> >> Has anyone recently looked at this? It definitely used to be stripped
> >> out. So something must have changed.
> >>
> >> Harbs
> >
> >
>

Re: traces not being stripped out

[Greg Dove <gr...@gmail.com>]

With the latest compiler updates, @royalesuppressexport should be working
again - I can now see the Vector stuff being dead-code-eliminated again if
it is not used, and 'org.apache.royale.utils.Language' and its members are
also no longer exported.
I did add some sdk level tests to monitor this feature, in the reflection
lib, so if anything changes and it no longer works, we should get to know
about it.
It was quite tricky to get this to work without changing the way it is
supported via ASDoc comment, because I think we are using those doc-comment
directives in ways that were perhaps never envisaged at the outset, but I
did find a sneaky way to do this that seems to work. It is using metadata
'under the hood' so it persists in the swf definitions inside a js swc,
which was the main problem that arose with the 'dynamic' nature of the
export instructions for Closure.

For the 'traces not being stripped out' issue, I did not look into that. If
you have a minimal example please open an issue and assign it to me, I will
look into it next weekend.


On Mon, Nov 1, 2021 at 2:28 AM Greg Dove <gr...@gmail.com> wrote:

> For the traces issue, I am not sure, but for Vector and others I believe
> that was related to compiler changes which meant that @royalesuppressexport
> no longer works, because the way things are being 'exported' changed.
>
> I will try to look into that as soon as I can. Josh and I had discussed
> this briefly, so this feature likely needs to use Metadata now iirc. I'll
> create an issue later today and assign to myself.
>
>
>
> On Mon, 1 Nov 2021, 12:46 am Harbs, <ha...@gmail.com> wrote:
>
>> I just noticed that traces are currently outputting in minified release
>> code.
>>
>> I also noticed code related to Vectors which should not be there. I
>> suspect that there’s more code that’s not being stripped as it should be.
>>
>> Has anyone recently looked at this? It definitely used to be stripped
>> out. So something must have changed.
>>
>> Harbs
>
>

Re: traces not being stripped out

[Greg Dove <gr...@gmail.com>]

For the traces issue, I am not sure, but for Vector and others I believe
that was related to compiler changes which meant that @royalesuppressexport
no longer works, because the way things are being 'exported' changed.

I will try to look into that as soon as I can. Josh and I had discussed
this briefly, so this feature likely needs to use Metadata now iirc. I'll
create an issue later today and assign to myself.



On Mon, 1 Nov 2021, 12:46 am Harbs, <ha...@gmail.com> wrote:

> I just noticed that traces are currently outputting in minified release
> code.
>
> I also noticed code related to Vectors which should not be there. I
> suspect that there’s more code that’s not being stripped as it should be.
>
> Has anyone recently looked at this? It definitely used to be stripped out.
> So something must have changed.
>
> Harbs