lse and you’ll still have this bug.
Thanks,
Justin
Hi,
I think we may be talking cross purposes here to be clear - IMO this is not an
optimisation issue this it’s a serious bug.
Justin
Hi,
Here’s the link to the latest failure onto build server and you can see it’s
typedef related.
https://builds.apache.org/job/FlexJS%20Framework%20(maven)/953/console
Thanks,
Justin
get:
error: patch failed: js/target/downloads/svg.js:401
error: js/target/downloads/svg.js: patch does not apply
So no maven and no ant involved and you still get an error.
Thanks,
Justin
t; It still still be possible to stage it from the tagged source RC or was there
> some other reason why it wasn’t staged?
Actually ignore this as it has been staged. [1]
Thanks,
Justin
1.
https://repository.apache.org/content/repositories/orgapacheflex-1020/org/apache/flex/flexjs/typedefs/
Hi,
> Again I’m in rush.
Sorry mean to say “Again I’m not in any rush."
Thanks,
Justin
Hi,
What do people feel about setting a code and branch coverage on all checkins?
Say 80% coverage of changes made?
Thanks,
Justin
s. As Dave suggested I will
document the changes to the rules in a wiki so we can discuss (if needed) and
implement.
Can you please revert the changes the to rules you have made.
Thanks,
Justin
hts?
Thanks,
Justin
Hi,
Looks like the issue is this, this has been running for 9+ hours and should
probably be killed [1]
May be a good idea to limit the time that job runs,it's usually fairly quick
right?
Thanks,
Justin
1. http://apacheflexbuild.cloudapp.net:8080/job/MD5Checker/15255/
single container and set the x and y
values to simulate padding. Complications occur were you have % height and
widths ie it not going to work as expected.
Thanks,
Justin
Hi,
Raised https://issues.apache.org/jira/browse/FLEX-35274 for this issue.
Justin
HI,
Raised https://issues.apache.org/jira/browse/FLEX-35272 for this issue.
Justin
s://www.google.com.au/webhp?sourceid=chrome-instant&ion=1&espv=2&ie=UTF-8#q=nick+nolte+mug+shot>
Thanks,
Justin
all content
on the screen ie if you are using visibility to swap panels make sure one is
displayed at all time.
Thanks,
Justin
Hi,
And the issue is each of the images is display inline-block this will by
default wrap unless you add white-space nowrap to the container.
Just do this to fix:
Hi,
"unrelated to a class would go" I mean.
Thanks,
Justin
ick up a
> lot of issues for you quickly because of that.
Thanks,
Justin
Hi,
Actually white-space: nowrap is on the Label not the Container but the question
is still anyone know why that was added?
Thanks,
Justin
Hi,
The vote passes with 3 +1 votes and no other votes.
Thanks to everyone who tested out this or any of the previous RCs.
Justin
Hi,
Here’s a more direct signup link if you need it:
https://the-asf.slack.com/signup <https://the-asf.slack.com/signup>
Thanks,
Justin
Hi,
I’ve looked through the code again and other than a couple of minor changes I
can't see any obvious way to to either reduce the lines of code or the logic
error Alex is referring to.
Someone care to enlighten me?
Thanks,
Justin
e.flex.html.accessories.*">
Thanks,
Justin
in the CSS. Only the first iFormat bead will have an effect.
Thanks,
Justin
Hi,
> No reason to be :-) Knowing what evacuates to true with == and false with !=
> is hard to remember and get right.
Evaluates was meant obviously. Autocorrect can sometime give amusing results.
thanks,
Justin
Gets optimised to:
test=function(){console.log("bar")};test();
So there’s no cost in this case for the initial assignment and the variable
doesn’t even exists in the final result. In more complex code you’ll probably
get something different.
Thanks,
Justin
ease is 7%.
Thanks,
Justin
Hi,
And nothing is broken in develop and a new release (after this one) is a long
way away so there’s no huge urgency here. I do intended to correct it once the
discussion has died down and PAYG is clearly documented.
Thanks,
Justin
ression that evaluates to false so just converting (b == false) to (!b)
would not be enough to solve the issue.
Thanks,
Justin
o see how often that pattern exists though.
From a quick glance (not testing)it exists in serval places in the existing
SDK.
Thanks,
Justin
Hi,
Perhaps a silly question but does there need to be a release of the typedef
project as well? Even if only for Maven?
Thanks,
Justin
Hi,
I notice falcon-flexTasks.jar is missing from the compiler/lib directory and
asdocs.jar and falcon-flexTasks.jar msiinsg form /js/lib. Should they be
included?
Thanks,
Justin
way or another in the next release.
Thanks,
Justin
ble?
Thanks,
Justin
Hi,
> I got it also. When I compiled typedefs it's gone.
Those error are from compiling typedefs, I’m using the latest develop branch.
Thanks,
Justin
!=
in the framework and implicit casting that goes on when when comparing
uninitialised variables.
Thanks,
Justin
ways
:-)
Thanks,
Justin
Hi,
> I did not see anything that would preclude a release, but the readme does not
> have anything that mentions Maven. The prerequisites only mention ANT and
> environment variables.
Falcon has this [1] should be easy enough to do similar.
Thanks,
Justin
1. https://github.com/ap
specific cases that perform better, please post any you find to the list.
Thanks,
Justin
"got here”)}
In this case both assignments are removed.
It even smart enough to turn this:
function test() {
var a = false;
var b = false;
a = true;
if (a && b) {
Console.log("got here");
}
}
Into nothing.
function test(){}
Thanks,
Justin
/maven2, releases=true,
snapshots=false)
[ERROR] Path to dependency:
[ERROR] 1) org.apache.flex.flexjs.framework:projects:pom:0.8.0-SNAPSHOT
Everything else looks fine so that’s the only think holding up my vote / that
and testing the release.
Thanks,
Justin
er.
Thanks,
Justin
nks,
Justin
correct
- LICENSE and NOTICE are fine
- all source files have ASF headers
- no unexpected binary files in release
- Can't compile from source (see discussion)
Also checked binary and that’s also fine re LICENSE and NOTICE etc ect
Thanks,
Justin
Hi,
Locally I have been using Maven for 4 months or so to build Falcon / FlexJS and
haven’t run into any major issues expect when the dual branch was first merged
and in the last couple of days so obviously something has changed recently that
stops this from working.
Thanks,
Justin
I’ll try 2.12.0 and see if that makes any difference.
> But not being able to release stuff to Maven-Central is reason enough for a
> -1 … sorry for that …
You'll need to do that in the [VOTE] thread for it to count.
Thanks,
Justin
from the tagged source release. [2] (and elsewhere)
Thanks,
Justin
1.
https://lists.apache.org/thread.html/47ad4369d92791280f8cdfb0d1e208cafb7d58860f0f743b54a76599@%3Cdev.flex.apache.org%3E
<https://lists.apache.org/thread.html/47ad4369d92791280f8cdfb0d1e208cafb7d58860f0f743b54a76599@%3
Hi,
> Justin - you did state that you consider this acceptable for this release and
> that we can clean it up for the next.
Yes I stated that in my vote and that would be acceptable.
Thanks,
Justin
cb9
Thanks,
Justin
1.
https://storage.googleapis.com/google-code-archive-downloads/v2/code.google.com/closureidl/svg.js
enkins build
system.
Thanks,
Justin
thanks,
Justin
Hi,
> Justin experienced this issue, maybe he will try fix with cache.
I can confirm that is the issue is as Greg explained. Like Greg once the issue
is fixed it actually hard to reproduce. I have some time today and will look
further into it and see if I can reliably reproduce and fix
ote to pass.
However a RM may still cancel a release if a serious issue is found and/or
thinks another RC is needed.
Thanks,
Justin
Hi,
> If you could once again reproduce this issue and try later whether switch
> off cache help that would be good.
>
> Go to: flex-typedefs\js\pom.xml and add to configuration tag in
> download-maven-plugin
>
> true
And I can confirm that this fixes the issue.
Thanks,
Justin
3rd party files have changed.
Thanks,
Justin
xt
I’ve not looked (been busy teaching a 3 day course) but that suggest a rat.txt
may of slipped into the release? Not an issue but probably also no need for it
to be there.
Thanks,
Justin
uld be used to calculate when copyright
expires (long after any of us are likely to still be alive). This has come up
in a couple of conversations on list(s) and at ApacheCon.
Thanks,
Justin
age it from the tagged source RC or was there
some other reason why it wasn’t staged?
Thanks,
Justin
n.
Thanks,
Justin
1. debugger/src/main/resources/flex/tools/debugger/cli/fdbhelp_*.txt
Hi,
I’m also seeing encoding difference between RC1 and RC2 in .as files [1][2][3]
I think it may only be in comments - look to be “smart" quotes.
Thanks,
Justin
1. frameworks/projects/Language/src/main/flex/QName.as
2. frameworks/projects/XML/src/main/flex/XML.as
3. frameworks/project
if (strpixels !== null && strpixels.indexOf('%') !== -1)
Thanks,
Justin
Hi,
> It broke when element was undefined.
So why not only change / revert that?
> There’s no reason to use strict equality so rampantly. (The performance gains
> are insignificant.)
As has previously been discussed on develop the performance gains are
significant.
Thanks,
Justin
/src/main/patch/google_maps.patch
flex-typedefs/jquery/src/main/patch/jquery.patch
flex-typedefs/js/src/main/patch/js.patch
I assume that this can be sorted out one way or another in the next release.
Thanks,
Justin
fixed and/or specific to our application.
But just a heads up if these issues turn out to be more wide spread and or
serious we may have to make a point release.
Thanks,
Justin
don’t seem too serious.
Thanks,
Justin
a cost of an extra byte in the
> download
No it doesn’t again you are forgetting the closure compiler optimises things
for us here and in the tests I’ve done so far there was no size increase and in
fact a size decrease.
Thanks,
Justin
type safety is generally
> enforced by the compiler
Expect were we’re comparing with undefined which we are currently doing for
just about every Object, String and Number in the SDK on the JS side.
Thanks,
Justin
Hi,
>> I got performance gains of 30-40%.
>
> Of what? Percentages when dealing with such cheap operators are generally
> meaningless.
Of total CPU time spent on application for a few minutes. Measured using the
Chrome profiler.
Thanks,
Justin
HI,
Point 1 may explain the issue clearer [1] there are situation we’re use of ==
and != is going to cause bugs.
Thanks,
Justin
1.
https://herringtondarkholme.github.io/2016/11/05/how-to-write-copy-paste-friendly-code/
<https://herringtondarkholme.github.io/2016/11/05/how-to-write-c
s from memory so 30-40% is a
significant improvement on that and noticeable by the user.
I'm out for the evening so will look up the email I sent to the dev list
tomorrow and post a link here.
Thanks,
Justin
HI,
Alternately you can compile the source release using maven and install locally
i.e. mvn clean install
Thanks,
Justin
be smaller by a fraction of a %.
I’ve also raised a JIRA re performance here. [2]
Thanks,
Justin
1.
https://herringtondarkholme.github.io/2016/11/05/how-to-write-copy-paste-friendly-code/
2. https://issues.apache.org/jira/browse/FLEX-35330
off it's likely you have less
performance and more bugs. I have a couple of these [1] and I’ll send you one
:-)
Thanks,
Justin
1. http://shop.evilmadscientist.com/productsmenu/62
false and act the same as in ActionScript rather than be undefined
as it is currently is in FlexJS.
Thanks,
Justin
hen initialised Boolean and Numbers will be undefined (in JS)
and default to the usual false and NaN is AS. This is the current behaviour of
the compiler.
Thanks,
Justin
and do some testing with the SDK
and a couple of applications. Again I’m in rush.
Thanks,
Justin
i.
> You made claims of much more significant improvements in a real app. If you
> have something that might back that up, I’d be interested in seeing…
Setting Boolean and Numbers to null gave a 10% increase in performance to our
application. I’ve not tested Objects yet.
Thanks,
Justin
re’s a performance gain and no downside and you can turn it off if you
want via a couple option what’s the harm in doing so?
Thanks,
Justin
iation so in my books that’s
reason enough to do it
> But initializing ints to null will generally get the same result as
> initializing them to 0…
Currently int's are initialised to 0 no one has suggested setting them to null.
Thanks,
Justin
Hi,
> You didn't get warnings when mixing undefined with Number or String?
No warnings are given.
The same code compiled using the FlexSDK also doesn’t give any warnings.
Thanks,
Justin
Hi,
I’m curious and certainly not a veto / asking you to revert this.
Is there any reason why this isn’t a BailoutDisableBead via PAYG concepts?
Thanks,
Justin
> On 25 Jun 2017, at 7:36 pm, ha...@apache.org wrote:
>
> Repository: flex-asjs
> Updated Branches:
> refs/head
mises the code and, add
compression on top of that and I’d be surprised if you see any significant
differences. I’ll try it out on the whole SDK and report back the results.
Thanks,
Justin
mvn test" locally on the develop branch the tests pass for me on OSX.
Thanks,
Justin
using this.
AllCSSValuesImpl is not performant and is unlikely to be used in a real
application so I don’t see much point in moving it there.
Thanks,
Justin
Hi,
> AllCSSValuesImpl - is not enough performant cause of lack of strict
> comparison ?
Just the large number of styles.When Harbs added it he said "It has a LOT of
styles, so it’s probably best to not use in as-is in deployed apps…”
Thanks,
Justin
Hi,
But you did agreed that the code in question should stay in CSSValuesImpl.
From you last email in that thread:
"I agree with Justin on this. I don’t see why numeric font weights is different
than backgroundImage for example. Not every app uses backgroundImage, but it
should work if
Hi,
> Considering we have AllCSSValuesImpl, I think that should be the default for
> apps which are expecting HTML CSS.
It’s not about HTTML CSS.
This code:
Will compile with no warnings but fail to work without the fix.
Thanks,
Justin
e.org/flexjs/basic">
With my fix it works as expected in JS i.e. the two lines are different sizes.
Thanks,
Justin
the boolean to false or not.
> Again, we don't have to all agree. We can teach the compiler to do
> different things at different points.
Sure.
Thanks,
Justin
our compiler currently supports that.
Thanks,
Justin
Hi,
The class is there I’m on the latest version of develop of all 3 repos and I’m
still getting that compile error. Did you actually try using that class /
compiling it?
Thanks,
Justin
HI,
> There are tests for both, and you DON’T get type conversions unless it’s
> required. Browsers are smarter than that.
See [1] and follow the path if x in undefined and y is a boolean and true.
Compare this with [2]
Thanks,
Justin
1. https://es5.github.io/#x11.9.3 <https://es5.
Hi,
Thanks for that Greg and good to see it started some conversation around this.
Thanks,
Justin
) define as successful here? We're currently working on an
application what does it need to do to help that?
Thanks,
Justin
1. https://en.wikipedia.org/wiki/Bus_factor
preferably a pointer
to the license in the LICENSE file [3], nothing would be added to NOTICE.
Including a link to SO may be helpful but links change over time.
Thanks,
Justin
1. https://www.apache.org/legal/resolved.html#cc-sa
2. https://www.apache.org/legal/resolved.html#category-b
3. http
Hi,
IMO downloading the binaries from an offical Apache page should be the
preferred link to hand out [1][2] even if a little more effort is needed to
make a working SDK.
Thanks,
Justin
[1] http://flex.apache.org/download-binaries.html
[2] http://flex.apache.org/download-flexjs.html
l. I know Chris looked deeper into the issues
here but not sure how far he got.
Thanks,
Justin
something.
Thanks,
Justin
1. https://builds.apache.org/analysis/overview?id=20942
e a few
weeks back.
> The singular code smell (Sonar says that flex smells very good indeed.)
Not sure where you seeing just 1 as I can see 13,100 and their are a varieties
of issues. Again some worth looking into and others not.
Thanks,
Justin
(10-20%) fairly quickly if new changes keep at the current TD rate.
Of course people have different options of what is clean and what isn’t and a
lot of what Sonar is reporting isn't going to matter so I would certainly
wouldn't take those numbers too literally.
Thanks,
Justin
1
1 - 100 of 6079 matches
Mail list logo
