On 10/2/15, 2:22 AM, "Christofer Dutz" <christofer.d...@c-ware.de> wrote:
>Common' ... JS ... seriously? What exactly would you be expecting from a >CI build for a JS project? Of course they aren't mainly thinking of CI >builds ... I'm talking about normal Software development. Every customer >I work for uses CI servers to ensure the quality of their software. > >I actually want to use it to make sure everything builds correctly, >automatically produce up to date maven snapshots of the parts that belong >to us and in general untangle the inter dependencies in our build. I am >used to working with complex builds and if I am having trouble working >out our current one, I bet most people wanting to contribute will be >having at least the same trouble. Most of them probably giving up. > >I want to change that. I want to lower the barrier for new contributions >and the main thing we need to address here is the build. Currently it >feels as if only 2-3 people are contributing to FlexJS as all ... I bet >this could be a lot more, if we reached out to the others. > >I know that you guys, sort of working on this full time, don't see the >problem, but at Apache it's "community over code", so we have to get the >community involved again. It isn’t that we don’t see a problem, it just a matter of priorities. The specific point here is whether folks will want to set up their own Jenkins builds of our source, not whether they use Jenkins in general. In your Enquiry/Poll, that is the only issue up for debate and that question is not in the list. I just don’t think that the majority of our committers are going to set up Jenkins builds of our source, especially since we will soon have them running again on my VM and maybe builds.a.o. I don’t think there is any disagreement that we want to make it easier for folks to get our source code and make changes. Making it easier to get the source and build it using Ant but not Jenkins was brought up a couple of months ago and I made two changes to the build scripts to do that. Has anybody tried it? The options are: 1) Follow the instructions here [1]. This sort of assumes you don’t already have a bunch of Apache Flex repo working copies you want to work with. Instead you will give a brand new empty folder and this script will bring down all of the right repos and build them in the correct order. 2) If you have already cloned flex-asjs, but don’t have any other repos cloned, you can run ‘ant all’ and it will bring down all the other repos and build them in the correct order. Some of our current committers have some collection of repos synced to various points in time and I have decided not to invest time in trying to figure that out for you. So yeah, there is this circularity between flex-falcon and flex-asjs, but if we can get the new script changes working it will not block people unless they are trying to get Jenkins builds of our source up and running. Since folks have complained when FlexJS via the Installer doesn’t work, I have to assume that the Installer is the primary way folks are getting FlexJS. Otherwise, we’d hear the complaint about the circularity more often and up the priority of resolving it. IMO, it isn’t worth debating how many people use Jenkins. And it isn’t worth complaining about the circularity exposed by Jenkins. It isn’t worth debating the merits of Maven. We want to attract the kinds of customers that use Maven. Instead, let’s implement or propose solutions. If you are building a compiler for a framework and both the compiler and framework are moving targets, how do you implement CI where the compiler should fail if the latest change to the compiler can’t compile the framework and the framework should fail if the compiler can’t compile the latest change to the source? -Alex [1] https://cwiki.apache.org/confluence/display/FLEX/FlexJS+Developer+Setup
