> I took a massive productivity hit when In-JVM dtests landed the codebase
These introduced new capabilities that were well understood as outcomes at the time. Here we are proposing replacing something that works fine, with something that works equivalently. The build file is not only consumed as a user, but also edited by everyone on the project. Most are familiar with it, can navigate it and edit it. This will be lost in any migration. Sure, like any migration it can be justified by its benefits. I just haven’t see anything specific besides that ant is “old” and people are “surprised” we use it. So what? It works. Sometimes things might be painful for some folk to do, but that will be true in any new build system, no? From: Paulo Motta <pauloricard...@gmail.com> Date: Thursday, 3 February 2022 at 11:53 To: Cassandra DEV <dev@cassandra.apache.org> Subject: Re: Build tool > I am productive today with ant. I will almost certainly take a productivity > hit sometime during and after the migration to maven, as will others. I took a massive productivity hit when In-JVM dtests landed the codebase, but this was not a consideration when that framework was added because it was perceived as a net gain for the project in the long term. In that case the productivity hit was definitely worth it, and I think in this case it will be. I personally don't think the productivity hit of adopting a new build tool will be very noticeable (nothing that you can't catch up in a couple of weeks), but in order to not block this effort on this feeling perhaps we can make reducing the productivity hit an explicit goal of this undertaking (ie. enumerate potential productivity hits and/or make the UX look as close as possible to the current). It could be helpful to come up with a concrete list of benefits and downsides of adopting a new build tool, to allow the community to decide whether the change is worth pursuing. Em qui., 3 de fev. de 2022 às 07:28, bened...@apache.org<mailto:bened...@apache.org> <bened...@apache.org<mailto:bened...@apache.org>> escreveu: > Aleksei has proven that he was able to deliver work of quality and to push > things forward. He is willing to try to tackle that work. I am not questioning his ability to deliver, I am questioning the value of burdening the project with this migration? I am productive today with ant. I will almost certainly take a productivity hit sometime during and after the migration to maven, as will others. From: Benjamin Lerer <ble...@apache.org<mailto:ble...@apache.org>> Date: Thursday, 3 February 2022 at 10:13 To: dev@cassandra.apache.org<mailto:dev@cassandra.apache.org> <dev@cassandra.apache.org<mailto:dev@cassandra.apache.org>> Subject: Re: Build tool I don’t have a super strong desire to stay with ant, I just have a desire not to unduly burden the project with unnecessary churn. Tooling changes can be quite painful. Aleksei has proven that he was able to deliver work of quality and to push things forward. He is willing to try to tackle that work. I am in favor of giving him the chance to do it. I fully agree that it is not a simple task but I am also convinced that other people might be interested in joining the effort. With regards to contributions, this is often brought up but the reality is the project has always struggled to bring in new ongoing contributors, in large part due to the barrier to entry of such a complex project (which has only grown as our expectations on patch quality have gone up). I struggle to believe that ANT is more than a rounding error on our efficacy here, since we have always struggled. I think we found some way around the barrier to entry :-). I will trigger another discussion about that. Le jeu. 3 févr. 2022 à 10:17, bened...@apache.org<mailto:bened...@apache.org> <bened...@apache.org<mailto:bened...@apache.org>> a écrit : I don’t have a super strong desire to stay with ant, I just have a desire not to unduly burden the project with unnecessary churn. Tooling changes can be quite painful. With regards to contributions, this is often brought up but the reality is the project has always struggled to bring in new ongoing contributors, in large part due to the barrier to entry of such a complex project (which has only grown as our expectations on patch quality have gone up). I struggle to believe that ANT is more than a rounding error on our efficacy here, since we have always struggled. If we’re struggling to actually use ant how we want that’s another matter, but it’s easy to forget how much just works for us with ant, and forget the things we will have pain with adopting a new build system. I have had more frustration with Gradle in a few months than I have with ant in a decade. I’m sure Maven is better, but I doubt it will be without issue. From: Benjamin Lerer <b.le...@gmail.com<mailto:b.le...@gmail.com>> Date: Thursday, 3 February 2022 at 09:03 To: dev@cassandra.apache.org<mailto:dev@cassandra.apache.org> <dev@cassandra.apache.org<mailto:dev@cassandra.apache.org>> Subject: Re: Build tool I think that there are 2 main issues (Aleksei can correct me): * ANT is pretty old and a lot of newcomers are unfamiliar with it and surprised by it. By consequence, it might slow down the on-boarding of newcomers which we want to make as smooth as possible. * Aleksei has been working on migrating our test to JUnit 5 and faced multiple issues with ANT. He provided five new features to the ANT project to fix the problems he encountered and some got rejected. I totally agree with your feeling that the current solution works for now and that staying with it is also a valid choice. I do like ANT. The question for me is really if ANT makes sense for the future of Cassandra. From the feedback I got, I start to doubt that it is the case. Le jeu. 3 févr. 2022 à 09:32, bened...@apache.org<mailto:bened...@apache.org> <bened...@apache.org<mailto:bened...@apache.org>> a écrit : I’m going to be a killjoy and once again query what value changing build system brings, that outweighs the disruption to current long-term contributors that can easily get things done today? At the very least there should be a ranked choice vote that includes today’s build system. From: Maulin Vasavada <maulin.vasav...@gmail.com<mailto:maulin.vasav...@gmail.com>> Date: Thursday, 3 February 2022 at 05:52 To: dev@cassandra.apache.org<mailto:dev@cassandra.apache.org> <dev@cassandra.apache.org<mailto:dev@cassandra.apache.org>> Subject: Re: Build tool Hi Aleksei I was thinking about the same - build tool. I have used both - Maven and Gradle. In my experience, while Gradle has a rich DSL and the corresponding power, with constant changes in Gradle across versions it is difficult to focus on the actual product (like Cassandra in this case) development. With Maven the learning is once and it doesn't change that much and one can focus on the actual product better. Of course, this is IMHO. +1 for using Maven. I would like to participate in the migration of the build tool if it needs more hands. Thanks Maulin On Wed, Feb 2, 2022 at 2:35 PM Aleksei Zotov <azotc...@apache.org<mailto:azotc...@apache.org>> wrote: Hi All, Some time ago I created https://issues.apache.org/jira/browse/CASSANDRA-17015 to migrate from ant to maven/gradle. Originally I was going to implement both, compare and pick the best in terms of project needs. However, now I feel it would be a significant overhead to try out both. Therefore, I'd like to make a collective decision on the build tool before starting any actual work. I saw on Slack (https://app.slack.com/client/T4S1WH2J3/CK23JSY2K/thread/CK23JSY2K-1643748908.929809) that many people prefer maven. I'm leaning towards maven as well. I guess we need to have a formal poll on the build tool since it is a significant part of the project. Please, suggest what the best way to proceed is. Should I just raise a vote for maven and just see if someone -1 in favor of gradle? PS: Please, bear in mind that Robert has already made some progress on gradle migration. I do not know how much is done there and whether he is willing to get it completed. On 2020/06/02 13:39:34 Robert Stupp wrote: > Yea - it's already in a pretty good state. > > Some work-in-progress-state is already available in either > https://github.com/snazy/cassandra/tree/tryout-gradle (or > https://github.com/snazy/cassandra/tree/tryout-gradle-dist-test with an > additional commit). > > I already use it on my machine for a bunch of things and it already > "feels bad" to go back to a branch without Gradle. > > I'll start a separate dev-ML thread with some more information in the > next days, because getting C* 4.0-beta released is a higher priority atm. > > On 6/1/20 2:41 AM, Joshua McKenzie wrote: > > Build tools are like religions, that's why. Or maybe cults. Or all > > Stockholm Syndrome creators? :) > > > > Robert Stupp has been noodling around with a gradle based build env for C* > > that'll live alongside ant. Not sure what the status is on that atm through. > > > > On Sun, May 31, 2020 at 3:16 PM Abhishek Singh > > <abh23...@gmail.com<mailto:abh23...@gmail.com>> wrote: > > > >> Hi All, > >> Hope you are doing well and are safe. > >> I just wanted to know why is the build still on ant and is there any plan > >> to migrate to a modern build tool? > >> > >> Regards, > >> Abhishek Singh > >> > -- > Robert Stupp > @snazy > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: > dev-unsubscr...@cassandra.apache.org<mailto:dev-unsubscr...@cassandra.apache.org> > For additional commands, e-mail: > dev-h...@cassandra.apache.org<mailto:dev-h...@cassandra.apache.org> > >
