Awesome! Staring from docker images and build up packages would be recommended so that you can get the idea about how it works. When you'd like to customize your own big data stack based on bigtop, you start to build your own docker images instead.
- Evans Olaf Flebbe <[email protected]> 於 2020年9月23日 週三 上午12:05寫道： > Hi Manfred, > > Great! > > AFAIK the most up to date information on how to (re)compile an package you > will find here > > https://cwiki.apache.org/confluence/display/BIGTOP/How+to+build+Bigtop-trunk > < > https://cwiki.apache.org/confluence/display/BIGTOP/How+to+build+Bigtop-trunk > > > > I am assuming you are targeting linux, not macos, right? > > IMO When developing it makes sense to compile „within" the provided docker > images or improving the docker images. > The CI’s job is to ensure changes do work on all platforms. > > I was using a macbook pro 16GB when I developed packages, so it is > possible, though not always optimal, since docker filesystem (volume mounts > from macos to docker guest) is buggy and slow. > > Running smoke tests on a mac is possible with workarounds. > > Needed workarounds depend very much on what you are trying to accomplish: > VMware Fusion, docker4mac or depending on CI are at least 3 options I would > recommend depending on usecase. > > Best > Olaf > > > > > > Am 22.09.2020 um 15:48 schrieb Manfred PAUL <[email protected]>: > > > > Hi, > > > > I would like to participate in developing bigtop. For that I am searching > > for some material (doc, wiki, video, etc) to get all the information > needed > > to build the bigtop distribution. > > > > All the information I found was for older versions of bigtop (1.1, 1.2 > and > > 1.3) but might still be up to date, so I am asking. > > > > Is written that a developer should use the pre-built docker images for > > building the distribution and not set up his or her own toolchain. Or > > should we use only a CI? Is it even possible to build everything on a > 16GB > > Macbook Pro? Etc. > > > > Thanks > > > > Manfred > >
