Minor bug in javascsript feature

[Randy Hudson <hu...@us.ibm.com>]

The feature "core.util.onload" makes a javascript call to 
gadgets.util.attachBrowserEvent(…), which is defined in the feature 
"core.util.event".

But, the onload feature doesn't have a dependency on the event feature, 
which means its javascript might execute before the attachBrowserEvent 
function has been declared.  This happens in our container.  I'm guessing 
it doesn't happen in shindig due to alphabetical sorting or some other 
shuffling that causes the event feature to come before onload.

The fix is pretty straightforward.

Just for grins, I'll attempt to attach a screenshot of a dev tool I wrote 
that displays the dependencies between some of shindig's features 
(modified to address this issue).

-Randy Hudson

Re: Minor bug in javascsript feature

[Randy Hudson <hu...@us.ibm.com>]

I notice my attachment was removed.  Here's a link to the screenshot:

https://dl.dropboxusercontent.com/u/14930473/OpenSocialDependencies.png

It creates a directed graph from the features, and then lays it out 
automatically.  I did this to understand which features we should disable, 
based on the features our container doesn't support.  (the colored lines 
are because the mouse was over the "rpc" feature)

-Randy



From:
Ryan Baxter <rb...@gmail.com>
To:
"dev@shindig.apache.org" <de...@shindig.apache.org>, 
Date:
06/19/2014 08:59 PM
Subject:
Re: Minor bug in javascsript feature



Hi Randy,

Looks like you are right after visually looking at the features.  I can
make this fix.

I am curious about the tool you wrote, would you mind sharing more 
details?

-Ryan


On Thu, Jun 19, 2014 at 5:48 PM, Randy Hudson <hu...@us.ibm.com> wrote:

> The feature "core.util.onload" makes a javascript call to
> gadgets.util.attachBrowserEvent(…), which is defined in the feature
> "core.util.event".
>
> But, the onload feature doesn't have a dependency on the event feature,
> which means its javascript might execute before the attachBrowserEvent
> function has been declared.  This happens in our container.  I'm 
guessing
> it doesn't happen in shindig due to alphabetical sorting or some other
> shuffling that causes the event feature to come before onload.
>
> The fix is pretty straightforward.
>
> Just for grins, I'll attempt to attach a screenshot of a dev tool I 
wrote
> that displays the dependencies between some of shindig's features 
(modified
> to address this issue).
>
> -Randy Hudson
>
>

Re: Minor bug in javascsript feature

[Ryan Baxter <rb...@gmail.com>]

Hi Randy,

Looks like you are right after visually looking at the features.  I can
make this fix.

I am curious about the tool you wrote, would you mind sharing more details?

-Ryan


On Thu, Jun 19, 2014 at 5:48 PM, Randy Hudson <hu...@us.ibm.com> wrote:

> The feature "core.util.onload" makes a javascript call to
> gadgets.util.attachBrowserEvent(…), which is defined in the feature
> "core.util.event".
>
> But, the onload feature doesn't have a dependency on the event feature,
> which means its javascript might execute before the attachBrowserEvent
> function has been declared.  This happens in our container.  I'm guessing
> it doesn't happen in shindig due to alphabetical sorting or some other
> shuffling that causes the event feature to come before onload.
>
> The fix is pretty straightforward.
>
> Just for grins, I'll attempt to attach a screenshot of a dev tool I wrote
> that displays the dependencies between some of shindig's features (modified
> to address this issue).
>
> -Randy Hudson
>
>