A ton of tests are good! Thank you for the effort. It really helps when
someone like me (who is just learning the codebase) wants to make changes.
I think your simple rule for handling member access expressions is
generally the right way to go. It just happens to need a special case when
a certain
I found the note at the bottom of the readme about the
compiler.tests/template-unittest.properties file. I had previously tried
adding a unittest.properties to define the environment variables, but I
guess I didn't get the details right at the time. Or maybe I didn't put it
in the right directory.
On Tue, Jan 12, 2016 at 7:58 PM, Josh Tynjala wrote:
> Mike,
>
> flash.utils.clearTimeout.length became a member access expression.
> IdentifierEmitter never used the fully-qualified name for a member access
> expression, so it was emitting as clearTimeout.length, without the package
> name.
>
>
Sounds good. What did you have to do to get the unit tests to run? It
might help the next person who tries.
Thanks,
-Alex
On 1/12/16, 11:36 PM, "Josh Tynjala" wrote:
>Before today, I could never get the unit tests to run on my local machine.
>However, now, I can do a full build of both flex-f
Before today, I could never get the unit tests to run on my local machine.
However, now, I can do a full build of both flex-falcon and flex-asjs
locally, so a broken build shouldn't happen as easily next time I need to
work on the compiler.
- Josh
On Tue, Jan 12, 2016 at 10:35 PM, Alex Harui wro
On 1/12/16, 10:29 PM, "Josh Tynjala" wrote:
>Good! Sorry for the trouble!
Well, if you looked closely, you'd see I made a change that broke the
build as well ;-) So I don't expect anyone to go forever without breaking
a build.
I try to always run "ant all" from the root of the flex-falcon re
Good! Sorry for the trouble!
- Josh
On Jan 12, 2016 9:23 PM, "Alex Harui" wrote:
> And the builds are passing. Thanks for sticking with it.
>
> -Alex
>
> On 1/12/16, 4:58 PM, "Josh Tynjala" wrote:
>
> >Mike,
> >
> >flash.utils.clearTimeout.length became a member access expression.
> >Identifie
And the builds are passing. Thanks for sticking with it.
-Alex
On 1/12/16, 4:58 PM, "Josh Tynjala" wrote:
>Mike,
>
>flash.utils.clearTimeout.length became a member access expression.
>IdentifierEmitter never used the fully-qualified name for a member access
>expression, so it was emitting as c
Mike,
flash.utils.clearTimeout.length became a member access expression.
IdentifierEmitter never used the fully-qualified name for a member access
expression, so it was emitting as clearTimeout.length, without the package
name.
See my commit and/or my reply to Alex for the explanation of how I fi
Thanks for the hint! I determined that it matters which side of the member
access expression the identifier appears on. If the identifier is on the
right of the expression, it should not be qualified.
- Josh
On Tue, Jan 12, 2016 at 3:22 PM, Alex Harui wrote:
> What I typically do in these situa
Josh, I have much experience with this identifier vrs. function call stuff
in the compiler. I pulled my hair out when initally implementing the
Identifier emitter and in the end made it as dumb as possible.
I was looking at your commits, and read this, can you exactly explain what
was failing , I'
What I typically do in these situations is examine the node tree and see
if I can see some difference that I can test for.
HTH,
-Alex
On 1/12/16, 3:17 PM, "Josh Tynjala" wrote:
>I'm running into some problem trying to fix this issue. I tried adding the
>following three tests to TestFlexJSExpres
I'm running into some problem trying to fix this issue. I tried adding the
following three tests to TestFlexJSExpressions:
@Test
public void testFunctionCallFullyQualified()
{
IFunctionNode node = (IFunctionNode) getNode(
"import goog.bind; public class B {public function b() {
goo
I discovered that when package-level and file-level functions or variables
are parented by a MemberAccessExpressionNode, the old code wasn't
outputting the proper fully-qualified name. This change was meant to handle
that special case. It looks like goog.bind is a package-level function,
though, so
Yep, I'm seeing this too. It could be the IdentifierEmitter changes. I
think the emitter code needs to check if it is already part of a
fully-qualified name, maybe by seeing if the IdentifierNode is parented by
a MemberAccessExpressionNode.
Josh, can you look into this? I think you can add a te
I updated falcon and flexjs this morning and ran into an issue. First
thought it was my problem then looked at the logs. When I build I get the
same issue as the warnings below and that causes a runtime failure.
[java] WARNING:
/Users/pent/dev/flex-asjs/examples/flexjs/MobileTrader/bin/js-debug/o
16 matches
Mail list logo
