Skip to content

Latest commit

 

History

History
354 lines (318 loc) · 12.2 KB

linux.markdown

File metadata and controls

354 lines (318 loc) · 12.2 KB
Raw

Flow: getting started (on Linux)

Much of the information in this getting-started is created for Ubuntu 18.04, so some of the operations may be different for your Operating System of choice.

Installation steps

  1. Environment configuration
  2. Backend
  3. Clone the repositories
  4. Install Haxe
  5. Install Neko
  6. Compile Flow itself
  7. Check it using flowcpp (C++ runner)
  8. Install fdb, the Flow debugger
  9. Check it using flowjs (Javascript in browser)
  10. Try it (Executed via apache, in browser)
  11. Tools
  12. Profiling

Sample install script

The .travis.yml file at the root level of the Flow repository is used on Travis-CI integration to install and configure Flow’s dependencies, build the Flow compiler, and run the flowunit tests suite, all on Ubuntu Linux. The commands in this file may be a useful guide for making the same happen on your own Linux system.

Environment configuration

All environment variables should be defined in ~/.profile. This is necessary since applications that are started from an application menu (i.e., not started from the command line; e.g., emacs) will not see environment variables defined in other files, such as ~/.bashrc. In some terminals, though, ~/.profile is not considered during startup, so as a compromise you can store variables in some file, which is source'd in both ~/.profile and ~/.bashrc

touch ~/.env
echo "source ~/.env" | tee -a ~/.bashrc ~/.profile

Also it can be useful to define environment variables in .xsessionrc. This way they will be enabled for any X session

Backend

MySQL

Set up MySQL and MySQL Workbench:

sudo apt install -y mysql-server mysql-client

Make root@localhost user with empty password:

sudo mysql --user="root" --execute="ALTER USER 'root'@'localhost' IDENTIFIED WITH mysql_native_password BY ''; FLUSH PRIVILEGES;"

Configure MySQL-server mode:

printf '[mysqld]
sql-mode=STRICT_ALL_TABLES,ERROR_FOR_DIVISION_BY_ZERO,NO_AUTO_CREATE_USER\n' | sudo tee -a /etc/mysql/my.cnf

More details on mysql setup can be found in flow9/doc/mysql.markdown

PHP5.6

Set up PHP:

sudo add-apt-repository ppa:ondrej/php
sudo apt update
sudo apt install -y php5.6 php5.6-mysql php-gettext php5.6-mbstring php-xdebug libapache2-mod-php5.6 php5.6-xml php5.6-zip php5.6-mcrypt
sudo update-alternatives --config php

Apache2

Install and configure apache2:

sudo apt install apache2
printf 'Alias "/todoapp" "/home/'$USER'/area9/todoapp/www2/"
<Directory /home/'$USER'/area9/todoapp/www2/>
     AllowOverride All
     Require local
</Directory>\n' | sudo tee /etc/apache2/conf-available/area9.conf
sudo a2enconf area9
sudo service apache2 restart

Note that todoapp is name for application in exercise 9, you can rename it as you wish.

Clone the repositories

If you haven't installed git yet, enter next commands:

sudo apt install -y git
git config --global user.email "<email for github account>"
git config --global user.name "<your full name>"
git config --global alias.hist "log --pretty=format:'%h %ad | %s%d [%an]' --graph --date=format:'%Y-%m-%d %H:%M:%S'"
git config --global push.default simple
git config --global pull.rebase true

In order not to enter authentication data each time, follow the instructions described in Connecting to GitHub with SSH.

Make area9 dir:

mkdir -p ~/area9 && cd ~/area9

Clone flow9 repos:

git clone ssh://git@github.com/area9innovation/flow9.git

Notice, that flow9 repo requires installed Git LFS. You have to reclone the flow9 repository after installing Git LFS, or use

git lfs pull

Install Haxe

Our build servers use haxe 3.2.1 and neko 2.0.0. Haxe 3.4.* should work.

A Linux installer is available here: Haxe download site

Note If you choose to use Linux package installer for Ubuntu, you may encounter following problem: add-apt-repository: command not found.
In such case, run following command: sudo apt-get update && sudo apt-get install software-properties-common

After running the installer, run this:

printf 'export HAXE_STD_PATH=/usr/share/haxe/std
export FLOW=$HOME/area9/flow9
export PATH=$FLOW/bin:$PATH\n' >> ~/.env

and run source ~/.env.

If you have followed the installation instructions for Linux Package installer, you have already set up Haxe library directory, so you don’t have to do the next step.

Set up your Haxe library directory:

cd ~
mkdir -p ~/haxelib/lib
haxelib setup ~/haxelib/lib

Then install the required libraries:

haxelib install format
haxelib install pixijs 4.7.1 #(or a newer edition)

Install Neko

Our build servers use haxe 3.2.1 and neko 2.0.0. Newer versions might work as well, but it is not guaranteed.

If the haxe installer above did not install Neko or installed a wrong version, you can do it manually either by downloading it from the official site

The [Neko download site](http://nekovm.org/download) offers both
32-bit and 64-bit binaries.  Make sure your haxe and neko match. Most
use the 32-bit version.

or via package manager like this:

sudo apt-get install neko

Run this:

printf 'export NEKOPATH=/usr/lib/x86_64-linux-gnu/neko
export PATH=$NEKOPATH:$PATH\n' >> ~/.env

and run source ~/.env.

NOTE: if you have error “Uncaught exception - load.c(237) : Failed to load library : std.ndll (dlopen(std.ndll, 1): image not found)” check your real neko path and put it in NEKOPATH, it can be /usr/local/lib/neko or another path.

Alexander Gavrilov offers:

there are two possible problems I remember with neko:

  1. neko executables work by including a small executable with the bytecode, and its 32-bitness must agree with libneko.so
  2. some distributions include weird ‘optimization’ things that strip all executables they see that they aren’t told to ignore, and that kills the bytecode specifically, point 2 is true for Fedora, and the weird thing is called prelink. the nekovm rpm included with the system installs a config into /etc/prelink.conf.d to stop that

LD_LIBRARY_PATH (Ubuntu)

The LD_LIBRARY_PATH environment variable must include the path to Neko. This is not done by explicitly setting this variable but by adding an application-specific file to the /etc/ld.so.conf.d directory like so:

Create the file /etc/ld.so.conf.d/flow.conf and add to it the path to Neko (/usr/bin/neko in Ubuntu). You can do this with:

echo '/usr/bin/neko' | sudo tee -a /etc/ld.so.conf.d/flow.conf

Run

sudo ldconfig

If you get the following messages

 /sbin/ldconfig.real: /usr/lib/nvidia-375/libEGL.so.1 is not a symbolic link
 /sbin/ldconfig.real: /usr/lib32/nvidia-375/libEGL.so.1 is not a symbolic link

then the following solution works

sudo mv /usr/lib/nvidia-375/libEGL.so.1 /usr/lib/nvidia-375/libEGL.so.1.org
sudo mv /usr/lib32/nvidia-375/libEGL.so.1 /usr/lib32/nvidia-375/libEGL.so.1.org
sudo ln -s /usr/lib/nvidia-375/libEGL.so.1 /usr/lib/nvidia-375/libEGL.so.375.39
sudo ln -s /usr/lib32/nvidia-375/libEGL.so.1 /usr/lib32/nvidia-375/libEGL.so.375.39

Note that LD_LIBRARY_PATH will not actually contain the path to Neko, but the system can now find libraries on that path nonetheless.

Compile Flow itself

Neko version of flow from the repo should be fine but if you wish to compile your own, do:

cd ~/area9/flow9/tools/flow
haxe FlowNeko.hxml

This should take a few seconds. If it takes more than a minute, very likely something is going wrong. The following file will be created or overwritten:

flow9/bin/flow.n

C++ runner (flowcpp)

Under linux it’s easier to compile yourself a binary instead of using the precompiled one. To do so, you should have Qt 5.12.0 set up by default Install required libraries:

sudo apt-get install zlib1g-dev libjpeg-dev libpng-dev -y
wget -q -O /tmp/libpng12.deb http://mirrors.kernel.org/ubuntu/pool/main/libp/libpng/libpng12-0_1.2.54-1ubuntu1_amd64.deb \
  && sudo dpkg -i /tmp/libpng12.deb \
  && rm /tmp/libpng12.deb

Download and setup QT 5.12.0 (or later):

wget https://download.qt.io/archive/qt/5.12/5.12.0/qt-opensource-linux-x64-5.12.0.run
chmod +x qt-opensource-linux-x64-5.12.0.run

Install into /opt/Qt5.12.0 and be sure to select all items to install! You might want to install into a different folder but it's important to create a symlink to the folder mentioned above and build with libraries statically linked from it if you plan on committing your QT byte runner to this repo.

./qt-opensource-linux-x64-5.12.0.run
rm qt-opensource-linux-x64-5.12.0.run
sudo apt install libpulse-dev libglu1-mesa-dev qtchooser -y
qtchooser -install qt512 ~/Qt/5.12.0/5.12.0/gcc_64/bin/qmake
echo "export QT_SELECT=qt512" >> ~/.env && source ~/.env

Clone asmjit repo:

cd $FLOW/platforms/common/cpp
git clone ssh://git@github.com/area9innovation/asmjit.git
cd asmjit
git checkout next

Build QtByteRunner:

cd $FLOW/platforms/qt
./build.sh # it can return with error 127, but that's expected

New QtByteRunner binary will appear in $FLOW/platforms/qt/bin/linux folder

Now you can run hello.flow using flowcpp:

cd ~/area9/flow9/
flowcpp sandbox/hello.flow

You should see:

Flow compiler (3rd generation)

Processing 'sandbox/hello' on server.
0.63s


Hello console

After that, the runner will hang, because the program doesn’t end with a call to quit(0), and the different runners aren’t consistent with each other about what happens when main() ends without an explicit quit call. You can ctrl-C out of it.

fdb, the Flow debugger

The C++ runner can also be used in debug mode, which activates a debugger with a GDB-like command line.

Wrapping the debugger in GNU Readline gives an improved experience with line editing and command history. If you haven’t already, do:

sudo apt-get install rlwrap

Then an example invocation to run the debugger is:

rlwrap flowcpp --debug sandbox/hello.flow

At the (fdb) prompt, you can enter commands such as step, next, and continue. The help command will show a list of all supported commands.

The debugger can also be used within an editor. Because it mimics GDB, it works with editors with gdb integration, for example Sublime Text package. See resources/sublimetext/readme.md for more details.

Try it (JavaScript in browser)

To compile Flow code to JavaScript, try e.g.

flowc1 sandbox/hello.flow js=www/hello.js

You should see:

Flow compiler (3rd generation)

Processing 'sandbox/hello' on server.

and flow9/www/hello.js should be created.

You might think to try the resulting JavaScript code with a command line tool like Node or SpiderMonkey, but that will break. So you’ll need to run it a browser.

A quick way to try that is:

xdg-open http://localhost/flow/flowjs.html?name=hello

Your browser should open and display a “Hello window” screen indefinitely and you look in the error console of your browser and see:

"Hello console"

Try it (Executed via apache, in browser)

Information on this topic can be found in platforms/qt/readme.md in section "Enabling fast-cgi in apache"

Tools

The auxiliary tools for Flow include a linter, a code formatter, and a refactoring tool.

They can be run directly using flow9/bin/lint.sh.

These tools are also used by the Sublime Text and Emacs integrations. (The editors also use flow9/bin/autocomplete.sh for autocompletion.)

Profiling

The instructions in development.markdown for using the Flow profiler should work fine on Mac as long as you have Java 8 installed.