I would like to know why you are using Google Docs and not the wiki.  IMO, 
unless folks on dev@ can follow discussion from comments in Google Docs, we 
should not be using Google Docs.

IMO, structured programming constructs in AS3 separate us from most other JS 
app-building choices.  We should be striving to make the use of structure more 
efficient and not encouraging the use of unstructured/dynamic constructs.   
Olaf had a good point about some of these features being about writing 
low-level JS.  Folks writing future-proof apps really shouldn't be calling 
JS/Browser classes directly.  We should be wrapping those things into proper 
abstractions.  IOW, if you instantiate a Blob, you are giving up portability.  
If you instantiate a BinaryData then you are not.  Underneath, we can do 
certain things to make the writing of the platform implementation by framework 
developers more efficient, but really, we shouldn't be encouraging users of 
Royale to also write to those low-level JS APIs.  Otherwise, folks are likely 
going to get in the same trouble they are with Flex.  No large app has shown up 
to be migrated that did not use flash.*.* classes and only used Flex classes 
that hid the low-level Flash stuff and where those migrating used Flash, that 
is really holding up their migration effort.  I don't think we should encourage 
users to use low-level JS APIs and have the same problem if we target something 
else someday.  It is pretty safe to assume that every future platform/runtime 
will support a binary array of data, but it unlikely they will support 
initializing that class with an object literal.

Also, IMO, Royale is more about a toolchain/workflow than a framework.  Really, 
we are trying to be framework agnostic.  We want to leverage structure in the 
language to allow the IDEs to be better than any JS-oriented IDE.

As such, it is worth considering whether some of these ideas, especially 
type-inferencing, should be done in the IDEs instead of the compiler.  The 
compiler will never be fast enough, so the more we make it think on every 
compile, the less productive you are.  Also look at really advanced IDEs for 
structured languages.  Do those JAVA IDEs support inferencing types from 
literals?  Why has the Java compiler decided not to inference types from 
literals?  At least, I don't think it does.  Usually there is a good reason 
behind these decisions that we should consider.

IMO, we want to put most of our energy into separating us from the pack not 
being more like the pack.

My 2 cents,
-Alex


On 1/11/19, 4:27 AM, "Harbs" <harbs.li...@gmail.com> wrote:

    I started a separate document with thoughts on typedefs. 
https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdocs.google.com%2Fdocument%2Fd%2F1gQTI08o5OJwawpludknyAX8oaa7quOZP9fridJWsKgg%2Fedit%3Fusp%3Dsharing&amp;data=02%7C01%7Caharui%40adobe.com%7Cffce206667d3415e498b08d677c015fc%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636828064220720562&amp;sdata=6RZNncFtAGeyT8hBBNXxq0G9yPI6yLDAJkUCY%2B908eg%3D&amp;reserved=0
 
<https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdocs.google.com%2Fdocument%2Fd%2F1gQTI08o5OJwawpludknyAX8oaa7quOZP9fridJWsKgg%2Fedit%3Fusp%3Dsharing&amp;data=02%7C01%7Caharui%40adobe.com%7Cffce206667d3415e498b08d677c015fc%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636828064220720562&amp;sdata=6RZNncFtAGeyT8hBBNXxq0G9yPI6yLDAJkUCY%2B908eg%3D&amp;reserved=0>
    
    The doc is editable. Feel free to add to it and make changes and 
corrections.
    
    It’s possible there is currently a way to specify source externs that I’m 
not aware of.
    
    My goal with my typedef thoughts are:
    1. Make typedefs easier to use.
    2. Improve type safety and easy of use with literal objects and JSON.
    3. Produce reliable output even after minification.
    
    Thanks,
    Harbs
    
    > On Jan 11, 2019, at 4:39 AM, Alex Harui <aha...@adobe.com.INVALID> wrote:
    > 
    > FWIW, we already support source externs.
    > 
    > On 1/10/19, 2:38 PM, "Harbs" <harbs.li...@gmail.com> wrote:
    > 
    >    I very much would like for AS3 to get an upgrade with features. That 
applies to improvements I the compiler output as well as truly new features in 
the language.
    > 
    >    I’ve started some discussion with Josh on the topic, and we started a 
Google doc to use to help figure out how we can incrementally improve things.
    > 
    >    The link to the Google Doc is here: 
https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdocs.google.com%2Fdocument%2Fd%2F188AAeny3y7bht9JbuE-RXIF_adZHP5uYj0--RANpGNM%2Fedit%3Fusp%3Dsharing&amp;data=02%7C01%7Caharui%40adobe.com%7Cffce206667d3415e498b08d677c015fc%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636828064220720562&amp;sdata=zgfXOasbwzADcuofrjyY%2FL1ppCUpucngpeVQ8vhOSwY%3D&amp;reserved=0
 
<https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdocs.google.com%2Fdocument%2Fd%2F188AAeny3y7bht9JbuE-RXIF_adZHP5uYj0--RANpGNM%2Fedit%3Fusp%3Dsharing&amp;data=02%7C01%7Caharui%40adobe.com%7Cffce206667d3415e498b08d677c015fc%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636828064220720562&amp;sdata=zgfXOasbwzADcuofrjyY%2FL1ppCUpucngpeVQ8vhOSwY%3D&amp;reserved=0>
    > 
    >    I enabled commenting. If someone wants edit access to the document, 
please let me know.
    > 
    >    Thanks,
    >    Harbs
    > 
    
    

Reply via email to