Hi,
I submitted a PR to the homebrew <http://brew.sh/> project (Mac’s unofficial package manager) to add a formula for installing Flink 1.0.2. https://github.com/Homebrew/homebrew-core/pull/968 <https://github.com/Homebrew/homebrew-core/pull/968> This will simplify installation on Mac. Here’s the typical flow: $ brew info apache-flink apache-flink: stable 1.0.2, HEAD Scalable batch and stream data processing https://flink.apache.org/ Not installed From: https://github.com/Homebrew/homebrew-core/blob/master/Formula/apache-flink.rb $ brew install apache-flink ==> Downloading https://www.apache.org/dyn/closer.lua?path=flink/flink-1.0.2/flink-1.0.2-bin-hadoop26-scala_2.10.tgz Already downloaded: /Library/Caches/Homebrew/apache-flink-1.0.2.tgz 🍺 /usr/local/Cellar/apache-flink/1.0.2: 88 files, 89.9M, built in 4 seconds $ flink --version Version: 1.0.2, Commit ID: d39af15 I made the somewhat arbitrary choice to use Scala 2.10 and Hadoop 2.6 for consistency with the apache-spark formula. In future the formula may be parameterized to allow the user to decide. I chose to add only the `flink`, `pyflink2.sh`, and `pyflink3.sh` commands to the system path. I felt that the names of the other bin scripts are too vague (`start-local.sh`, etc) to be on the system path. I considered `flink-daemon` but was unsure it is a documented tool. There’s still a chance to change the PR before it is merged. Of course we can change it further with 1.0.3. Enjoy, Eron Wright |
Update: I filed FLINK-3883 to improve the names of the shell scripts for inclusion on the system path. Pending a discussion on that, I adjusted the formula PR to place only the `flink` command on the path (not the pyflink commands).
Reflecting a bit, I don’t think the various server scripts (flink-daemon.sh, start-cluster.sh, etc.) belong on the system path. Eron > On May 8, 2016, at 1:38 PM, Wright, Eron <[hidden email]> wrote: > > Hi, > > I submitted a PR to the homebrew <http://brew.sh/> project (Mac’s unofficial package manager) to add a formula for installing Flink 1.0.2. > https://github.com/Homebrew/homebrew-core/pull/968 <https://github.com/Homebrew/homebrew-core/pull/968> > > This will simplify installation on Mac. Here’s the typical flow: > > $ brew info apache-flink > apache-flink: stable 1.0.2, HEAD > Scalable batch and stream data processing > https://flink.apache.org/ > Not installed > From: https://github.com/Homebrew/homebrew-core/blob/master/Formula/apache-flink.rb > > $ brew install apache-flink > ==> Downloading https://www.apache.org/dyn/closer.lua?path=flink/flink-1.0.2/flink-1.0.2-bin-hadoop26-scala_2.10.tgz > Already downloaded: /Library/Caches/Homebrew/apache-flink-1.0.2.tgz > 🍺 /usr/local/Cellar/apache-flink/1.0.2: 88 files, 89.9M, built in 4 seconds > > $ flink --version > Version: 1.0.2, Commit ID: d39af15 > > I made the somewhat arbitrary choice to use Scala 2.10 and Hadoop 2.6 for consistency with the apache-spark formula. In future the formula may be parameterized to allow the user to decide. > > I chose to add only the `flink`, `pyflink2.sh`, and `pyflink3.sh` commands to the system path. I felt that the names of the other bin scripts are too vague (`start-local.sh`, etc) to be on the system path. I considered `flink-daemon` but was unsure it is a documented tool. > > There’s still a chance to change the PR before it is merged. Of course we can change it further with 1.0.3. > > Enjoy, > Eron Wright |
Hi Eron!
Thanks for the brew formula! I agree that flink-daemon doesn't belong in the system path. I think it would make sense to prepend all the Flink startup scripts with "flink-". That way they are easily discoverable. Cheers, Max On Mon, May 9, 2016 at 1:58 AM, Wright, Eron <[hidden email]> wrote: > Update: I filed FLINK-3883 to improve the names of the shell scripts for inclusion on the system path. Pending a discussion on that, I adjusted the formula PR to place only the `flink` command on the path (not the pyflink commands). > > Reflecting a bit, I don’t think the various server scripts (flink-daemon.sh, start-cluster.sh, etc.) belong on the system path. > > Eron > > >> On May 8, 2016, at 1:38 PM, Wright, Eron <[hidden email]> wrote: >> >> Hi, >> >> I submitted a PR to the homebrew <http://brew.sh/> project (Mac’s unofficial package manager) to add a formula for installing Flink 1.0.2. >> https://github.com/Homebrew/homebrew-core/pull/968 <https://github.com/Homebrew/homebrew-core/pull/968> >> >> This will simplify installation on Mac. Here’s the typical flow: >> >> $ brew info apache-flink >> apache-flink: stable 1.0.2, HEAD >> Scalable batch and stream data processing >> https://flink.apache.org/ >> Not installed >> From: https://github.com/Homebrew/homebrew-core/blob/master/Formula/apache-flink.rb >> >> $ brew install apache-flink >> ==> Downloading https://www.apache.org/dyn/closer.lua?path=flink/flink-1.0.2/flink-1.0.2-bin-hadoop26-scala_2.10.tgz >> Already downloaded: /Library/Caches/Homebrew/apache-flink-1.0.2.tgz >> 🍺 /usr/local/Cellar/apache-flink/1.0.2: 88 files, 89.9M, built in 4 seconds >> >> $ flink --version >> Version: 1.0.2, Commit ID: d39af15 >> >> I made the somewhat arbitrary choice to use Scala 2.10 and Hadoop 2.6 for consistency with the apache-spark formula. In future the formula may be parameterized to allow the user to decide. >> >> I chose to add only the `flink`, `pyflink2.sh`, and `pyflink3.sh` commands to the system path. I felt that the names of the other bin scripts are too vague (`start-local.sh`, etc) to be on the system path. I considered `flink-daemon` but was unsure it is a documented tool. >> >> There’s still a chance to change the PR before it is merged. Of course we can change it further with 1.0.3. >> >> Enjoy, >> Eron Wright > |
Free forum by Nabble | Edit this page |