Hi Alex, you're right, and Chris is aware about to make Ant continue working. If I understand correctly, he worked as well in ANT scripts to clean and streamline as possible, so that's part of the changes :)
As you said, we still need to wait for his end this current work, to see more, but I want to announce here so you all know about it. Hope the changes come soon in the next days and we could finally see the benefits. As well hope the changes will more easy to deal with the current CI release steps failing. In the end, I think we all want lower the complexity of building and release, while maintain granularity, so all this changes go in that direction :) Thanks! Carlos El mié., 4 dic. 2019 a las 18:10, Alex Harui (<[email protected]>) escribió: > Sounds promising. I guess we'll see when more actual work is done. > > I still don't get the part about compiler dependencies. I thought it > would take yet another repo/submodule to handle the current utils profile. > > Keep in mind that whoever is making these changes is responsible for > making sure the Ant build still runs and the CI release steps still run. I > am nervous because the last set of changes broke many of the CI release > steps. > > My 2 cents, > -Alex > > > > On 12/4/19, 1:54 AM, "Carlos Rovira" <[email protected]> wrote: > > Hi, > > will try to elaborate and complement over I wrote yesterday: > > The plan is to have a 4th repo, and there have 3 Sub Modules, each one > as > we have now (compiler, typedefs and framework). > Ant will be work as always (calling "ant all" from framework will > build as > usual, and other targets should work as usual). > Each Maven sub module will work as always too separately if we want, > since > some people want to use just compiler (Kenny?, Josh?, others we even > don't > know?,...), so although we get a single monolithic repo, we still have > the > granularity to choose what parts we want to release when we want to > start > doing so. In other words, Maven will be prepared to do so, but, maybe > ANT > will need work if even now it does not consider that, but that is > something independent of doing the proposed change. > > What Chris is refactoring is the maven plugin to avoid dependencies, > that > is what Alex state that will be afraid to have more. This is just the > opposite. We have now dependencies that want to remove (the compiler > dependency. will be removed to make them more decoupled). So I think > that > goes in the right direction that Alex was worried about. > > Another point is to be able to release each Sub Module on its own, and > that > would be as well possible. > The Good point is that after this change we will be able to build all > with > maven with just a single command instead of do the same command in > three > different repos. > > For releases still all have to be released manually separately, hovever > another good catch is that Chris is defaulting the properties that > define > the compiler version an typedef version to project version. And that > can be > overridden too. But that will be chained making it more easy since > we're > using the submodules and maven as intended by those tools. > > Another collateral point is that Chris is greatly cleaning up the > compiler > plugin configuration, that will streamline all the process involved. > > Let me know if this clarifies the proposal and the enhancement we want > to > do, and if is ok for you all. > > Thanks! > > Carlos > > > El mar., 3 dic. 2019 a las 20:52, Alex Harui (<[email protected] > >) > escribió: > > > From a release standpoint I'm concerned about the scalability of a > single > > monolithic release. I would rather we head towards smaller subset > > releases. Even the Ant release we do now is cumbersome. But from a > "get > > the code and build it" standpoint, a top-level repo that simplifies > Maven > > instructions seems worth it. > > > > -Alex > > > > On 12/3/19, 10:47 AM, "Carlos Rovira" <[email protected]> > wrote: > > > > Hi, > > > > @Harbs <[email protected]> you are right. the proposal is not > > disruptive, > > tries to streamline what we have. After reading the responses it > seems > > my > > explanation could be bad, since all seems to be misunderstood. > > > > Will try to elaborate more soon :) > > > > -- > > Carlos Rovira > > > https://nam04.safelinks.protection.outlook.com/?url=http%3A%2F%2Fabout.me%2Fcarlosrovira&data=02%7C01%7Caharui%40adobe.com%7C93d1b81fe5e841a308f808d7789ff26c%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C637110500676586885&sdata=yy9qC9JUoV2pcYLEsX%2BEOxNPnuui%2FpzDIADS4nXowqs%3D&reserved=0 > > > > > > > > > > > -- Carlos Rovira http://about.me/carlosrovira
