You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@flex.apache.org by Jun Heider <ju...@realeyes.com> on 2012/02/21 01:30:01 UTC
Committing to Trunk
Now that Carol has gotten the trunk populated, I was wondering if we need to wait until the Mustella tests are in before committing to trunk? The reason I ask is I know that there's been some patches submitted to the JIRA tickets and it would be nice to start getting work committed.
On another note, have we decided on a standard process for committing to trunk? For instance is +1 voting on the ML the Apache version of code reviews - I hope not since that seems like it'd be too slow and cumbersome. Also, without waiting for Mustella how do we test code before committing?
If there is a process already I'm sure it's documented somewhere and I'm just missing it, but I figured I'd start this dialog now before the backlog of patches gets too large to maintain.
Thanks.
Re: Committing to Trunk
Posted by Jun Heider <ju...@realeyes.com>.
On Feb 20, 2012, at 6:02 PM, Bertrand Delacretaz wrote:
> On Mon, Feb 20, 2012 at 4:42 PM, Jun Heider <ju...@realeyes.com> wrote:
>>
>> Is the JIRA hooked up to the SVN repo? For instance being able to add a
>> "RE: {Ticket Number}" to the commit message?
>
> Yes, if you start a commit message with FLEX-1234 the corresponding
> comments are generated in the FLEX-1234 issue.
>
> -Bertrand
Nice, good to know!
Re: Committing to Trunk
Posted by Bertrand Delacretaz <bd...@apache.org>.
On Mon, Feb 20, 2012 at 4:42 PM, Jun Heider <ju...@realeyes.com> wrote:
>
> On Feb 20, 2012, at 5:38 PM, Bertrand Delacretaz wrote:
> ...I think personally, with JIRA being around my preference would be a Review-Then-Commit.
> Is the JIRA hooked up to the SVN repo? For instance being able to add a
> "RE: {Ticket Number}" to the commit message?
Yes, if you start a commit message with FLEX-1234 the corresponding
comments are generated in the FLEX-1234 issue.
-Bertrand
Re: Committing to Trunk
Posted by Jun Heider <ju...@realeyes.com>.
On Feb 20, 2012, at 5:38 PM, Bertrand Delacretaz wrote:
> On Mon, Feb 20, 2012 at 4:30 PM, Jun Heider <ju...@realeyes.com> wrote:
>> ...have we decided on a standard process for committing to trunk? For instance is +1 voting on the ML the
>> Apache version of code reviews - I hope not since that seems like it'd be too slow and cumbersome...
>
> The most common way in Apache projects is CTR, Commit-Then-Review: you
> commit, and someone can complain if they don't like it (including
> vetos if really needed, described at
> http://www.apache.org/foundation/voting.html). That's by far my
> favorite, if people are paying attention (subscribed to the
> flex-commits list etc.) bad things are caught quickly, and people try
> to be careful to avoid looking stupid too often.
>
> The PPMC might decide to use RTC, Review-Then-Commit for some critical
> parts if needed, where a suggested patch (attached to a JIRA issue)
> needs to be reviewed before being applied. A committer can also always
> go that route and ask for review if unsure about a particular patch.
>
> -Bertrand
I think personally, with JIRA being around my preference would be a Review-Then-Commit. Is the JIRA hooked up to the SVN repo? For instance being able to add a "RE: {Ticket Number}" to the commit message?
Re: Committing to Trunk
Posted by Bertrand Delacretaz <bd...@apache.org>.
On Mon, Feb 20, 2012 at 4:30 PM, Jun Heider <ju...@realeyes.com> wrote:
> ...have we decided on a standard process for committing to trunk? For instance is +1 voting on the ML the
> Apache version of code reviews - I hope not since that seems like it'd be too slow and cumbersome...
The most common way in Apache projects is CTR, Commit-Then-Review: you
commit, and someone can complain if they don't like it (including
vetos if really needed, described at
http://www.apache.org/foundation/voting.html). That's by far my
favorite, if people are paying attention (subscribed to the
flex-commits list etc.) bad things are caught quickly, and people try
to be careful to avoid looking stupid too often.
The PPMC might decide to use RTC, Review-Then-Commit for some critical
parts if needed, where a suggested patch (attached to a JIRA issue)
needs to be reviewed before being applied. A committer can also always
go that route and ask for review if unsure about a particular patch.
-Bertrand
Re: Committing to Trunk
Posted by Jun Heider <ju...@realeyes.com>.
On Feb 20, 2012, at 6:15 PM, Justin Mclean wrote:
>> Now that Carol has gotten the trunk populated, I was wondering if we need to wait until the Mustella tests are in before committing to trunk?
> Just as an aside I've selected those issues as it was something that could be easily verified without Mustella (ie no real framework code changes). I'm not sure that Mustella would actually be able to test those patches in any meaningful way. Until we get our hands on it hard to say.
Yeah I noticed that the issues you worked on were locale issues/enhancements. Your statement makes sense to me regarding verifiable vs. non-verifiable issues. I guess I was just wanting to start a general discussion around committing, and I think whether we have one route to commit vs. multiple it would be nice to get those routes fleshed out so we can document them somewhere.
>
>> On another note, have we decided on a standard process for committing to trunk?
> Well for lower risk smaller changes I'm all for CTR, for larger higher risk changes RTC may be more suitable. I'm sure anyone who's a committers can use their judgment on a case by case basis?
This makes sense too. If we get the commit processes documented I think that it would be just fine to have a couple paths to trunk. (Low risk/high risk)
Thanks for the input!
Re: Committing to Trunk
Posted by Jun Heider <ju...@realeyes.com>.
On Feb 20, 2012, at 6:29 PM, Alex Harui wrote:
> On 2/20/12 5:15 PM, "Justin Mclean" <ju...@classsoftware.com> wrote:
>>
>>> On another note, have we decided on a standard process for committing to
>>> trunk?
>> Well for lower risk smaller changes I'm all for CTR, for larger higher risk
>> changes RTC may be more suitable. I'm sure anyone who's a committers can use
>> their judgment on a case by case basis?
> I think the committers should use CTR. And you can always ask for a review
> up front based on your judgement. Folks who aren't committers will submit
> patches and find a way to get them reviewed by a committer.
This makes sense. I mean we have a repo. If someone makes a commit boo-boo we can always revert need be. So cool, thanks Alex.
Re: Committing to Trunk
Posted by Justin Mclean <ju...@classsoftware.com>.
Hi,
> Sure. If there is no test for it, then there is no way you can "break the
> build" by checking it in.
Not having access to Mustella I can't really say if it tests for something or not I can only guess. Hopefully it will be donated soon. :-)
> Since you are not a committer yet, we just need to find a committer
> to review it
Any PPMC members who are English could review it as there would be minimal differences between en_GB and en_AU despite what you see/hear in films :-) I'll also see if I can get some locals to have a look and provide feedback. If you look at the diffs from en_US which I provided in JIRA there's not a lot of changes.
> Yes, I want to continue to use ANT
So does anyone know what we need to do to set this up in Apache infrastructure?
Thanks,
Justin
Re: Committing to Trunk
Posted by Alex Harui <ah...@adobe.com>.
On 2/20/12 7:33 PM, "Justin Mclean" <ju...@classsoftware.com> wrote:
> Hi,
>
>> Definitely no need to wait to commit changes for which there are no Mustella
>> tests. However, I thought we were going to try to cut a release that was
>> close to "parity" first.
> Can we do that without Mustella? or the compiler?
Sure. If there is no test for it, then there is no way you can "break the
build" by checking it in. I'm thinking specifically about your locale
patch. Since you are not a committer yet, we just need to find a committer
to review it (or in this case, oversee the review since I don't think there
are other Australians on the PPMC).
>
>
> We have an ant build script that works checked into SVN (build_framwork.xml)
> why not use that for now until something better is decided?
Yes, I want to continue to use ANT, but our continous integration at Adobe
was using Cruise Control to run that ANT script. There was a past
discussion on Apache-owned CI technology. I think we should use one of them
to run our ANT script. I'm not in favor of switching away from ANT to Maven
or other things like it at this time.
Making the releases work with Maven is a different effort that everyone
wants to see.
--
Alex Harui
Flex SDK Team
Adobe Systems, Inc.
http://blogs.adobe.com/aharui
Re: Committing to Trunk
Posted by Justin Mclean <ju...@classsoftware.com>.
Hi,
> Definitely no need to wait to commit changes for which there are no Mustella
> tests. However, I thought we were going to try to cut a release that was
> close to "parity" first.
Can we do that without Mustella? or the compiler? I guess the question is do we wait around until we have everything or try and continue the best we can without those missing parts? Currently I assume they are held up in legals and it's unsure when exactly they will be donated? There's nothing stopping us from making a release from the initial version after code has been submitted into SVN as you can build from any SVN revision you care to select.
Personally I wouldn't mind if the initial release had a few bug fixes and minor improvements as it gives a reason for using the Apache SDK over the Adobe Flex 4.6 SDK. If we do have patches in SVN it means we can make a second release shortly after the initial one, once we have the build and testing process sorted out.
> I think we have agreement on continuous integration, it is just a matter of
> deciding what technology to use and getting it done.
We have an ant build script that works checked into SVN (build_framwork.xml) why not use that for now until something better is decided?
Thanks,
Justin
Re: Committing to Trunk
Posted by Alex Harui <ah...@adobe.com>.
On 2/20/12 5:15 PM, "Justin Mclean" <ju...@classsoftware.com> wrote:
> Hi there.
>
>> Now that Carol has gotten the trunk populated, I was wondering if we need to
>> wait until the Mustella tests are in before committing to trunk?
> Just as an aside I've selected those issues as it was something that could be
> easily verified without Mustella (ie no real framework code changes). I'm not
> sure that Mustella would actually be able to test those patches in any
> meaningful way. Until we get our hands on it hard to say.
Definitely no need to wait to commit changes for which there are no Mustella
tests. However, I thought we were going to try to cut a release that was
close to "parity" first.
>
>> On another note, have we decided on a standard process for committing to
>> trunk?
> Well for lower risk smaller changes I'm all for CTR, for larger higher risk
> changes RTC may be more suitable. I'm sure anyone who's a committers can use
> their judgment on a case by case basis?
I think the committers should use CTR. And you can always ask for a review
up front based on your judgement. Folks who aren't committers will submit
patches and find a way to get them reviewed by a committer.
>
> We also need to have a discussion about how we get a nightly build set up and
> if we build the framework when changes are submitted to SVN?
I think we have agreement on continuous integration, it is just a matter of
deciding what technology to use and getting it done.
--
Alex Harui
Flex SDK Team
Adobe Systems, Inc.
http://blogs.adobe.com/aharui
Re: Committing to Trunk
Posted by Justin Mclean <ju...@classsoftware.com>.
Hi there.
> Now that Carol has gotten the trunk populated, I was wondering if we need to wait until the Mustella tests are in before committing to trunk?
Just as an aside I've selected those issues as it was something that could be easily verified without Mustella (ie no real framework code changes). I'm not sure that Mustella would actually be able to test those patches in any meaningful way. Until we get our hands on it hard to say.
> On another note, have we decided on a standard process for committing to trunk?
Well for lower risk smaller changes I'm all for CTR, for larger higher risk changes RTC may be more suitable. I'm sure anyone who's a committers can use their judgment on a case by case basis?
We also need to have a discussion about how we get a nightly build set up and if we build the framework when changes are submitted to SVN?
Thanks,
Justin
Re: Committing to Trunk
Posted by Jun Heider <ju...@realeyes.com>.
On Feb 20, 2012, at 5:37 PM, Michael A. Labriola wrote:
>
> Jun,
>
> Thanks for asking. I was looking at the same thing today for Justin's patch.
>
> Mike
>
Yep, Justin's patches are the ones I noticed. Thanks Justin! :)
RE: Committing to Trunk
Posted by "Michael A. Labriola" <la...@digitalprimates.net>.
>Now that Carol has gotten the trunk populated, I was wondering if we need to wait until the Mustella tests are in before committing to trunk? The reason I ask is I know that there's been some patches submitted to the JIRA >tickets and it would be nice to start getting work committed.
>
>On another note, have we decided on a standard process for committing to trunk? For instance is +1 voting on the ML the Apache version of code reviews - I hope not since that seems like it'd be too slow and cumbersome. Also, >without waiting for Mustella how do we test code before committing?
>
>If there is a process already I'm sure it's documented somewhere and I'm just missing it, but I figured I'd start this dialog now before the backlog of patches gets too large to maintain.
>
Jun,
Thanks for asking. I was looking at the same thing today for Justin's patch.
Mike