Skip to content

Latest commit

 

History

History
441 lines (317 loc) · 19.2 KB

Workflow.md

File metadata and controls

441 lines (317 loc) · 19.2 KB

Workflow

This guide will walk through the different workflows recommended for developing Vireo. It covers setting up your own fork with continuous integration (CI) server connections as well as commands for developing locally, keeping the source up to date, and submitting changes.

Getting started

Create a fork

Vireo relies on the fork-pull model for developer contributions. To begin development on Vireo make a GitHub account and fork the ni/VireoSDK repository.

Your fork will be available online at github.com/YOUR_USER_NAME/VireoSDK

CI Server

Vireo uses GitHub Actions to build and test the Vireo source code automatically whenever changes are made. You are encouraged to push changes to your fork when you have changes and to make branches for experiments. The fork is your playground for development and experimentation.

From GitHub you can look at the Actions tab to see the status of the CI for your branches.

Clone fork

With the above completed you have now taken the ni/VireoSDK repository (the upstream repository) and made a fork on GitHub as YOUR_USER_NAME/VireoSDK (the origin repository).

In order to do development in your fork you need to make a clone of the fork on your development machine.

An example would be to navigate to a folder on disk and run:

git clone https://github.com/YOUR_USER_NAME/VireoSDK.git

This will create a folder named VireoSDK with the associated Vireo source code.

Verifying the clone

Once the source has been cloned into the VireoSDK folder you can navigate into the folder and run:

> git remote -v
origin  https://github.com/YOUR_USER_NAME/VireoSDK.git (fetch)
origin  https://github.com/YOUR_USER_NAME/VireoSDK.git (push)
upstream https://github.com/ni/VireoSDK.git (fetch)
upstream https://github.com/ni/VireoSDK.git (push)

If you used GitHub Desktop to clone the fork you should see both the remotes listed above.

If you used the git command-line tool to clone the fork you may need to add the upstream remote to your local clone:

git remote add upstream https://github.com/ni/VireoSDK.git

Development workflow

In the fork-pull workflow new development is performed in small short-lived branches in your fork and a pull request is submitted to have those changes pulled in upstream. Generally, you should not work directly in the master branch of your fork but instead create a branch from master for development.

When changes are pulled into upstream from your branch they can be either merged directly or modified (squashed or rebased). The ability of pull request merges to modify commits is one reason long-lived branches are undesireable as maintenance is required to avoid diverging from master.

Creating a branch

To create a branch first check the branch you are currently on by running:

> git status
On branch master
Your branch is up to date with 'origin/master'.

nothing to commit, working tree clean

It is recommended that new work be performed by branching from master. To change branches to the master branch run:

> git checkout master
Switched to branch 'master'
Your branch is up to date with 'origin/master'.

Then make a branch from the current branch named awesome_branch:

> git checkout -b awesome_branch
Switched to a new branch 'awesome_branch'

To push the new branch and any local changes to your fork the first-time:

> git push -u origin awesome_branch
Total 0 (delta 0), reused 0 (delta 0)
To https://github.com/YOUR_USER_NAME/VireoSDK.git
 * [new branch]      awesome_branch -> awesome_branch
Branch 'awesome_branch' set up to track remote branch 'awesome_branch' from 'origin'.

To push local changes in subsequent updates:

git push

You are encouraged to push changes frequently (at least once a day if not more often) to your fork. This allows other developers to see your code as well as backs up the code on the GitHub servers. In addition, pushing code to a fork allows the CI servers to run tests on your code, potentially helping find breaking changes sooner.

Feel free to create branches for new features, fixing issues, experiments, etc. and have them in your fork. Your fork is your workspace to develop freely and experiment. Just make sure to use descriptive and concise names for your branches so others can identify them.

Bringing branches up to date

Changes that happen upstream are not automatically pulled into your fork or your local clone.

One concept to be aware of is that git stores a pool of all the changes (git objects) from the remotes that you are connected to. This pool of objects are not kept up to date automatically. To fetch all the latest objects from all of your remotes (while also deleting branches that are deleted in remotes) run:

git remote update -p

Bringing master up to date

Based on the recommended workflow the master branch of your local clone should not have any commits that diverge from upstream. Because there are no commits that diverge from upstream it is possible to fast forward merge your local master.

To do this first verify you are on master with git status or change to master with git checkout master.

Then bring your local master up to date by running:

git merge --ff-only upstream/master

If you get an error that means you have unexpected commits on your master branch that should be removed.

If everything succeeds then you can go ahead and push your updated local master to your fork:

git push

Rebasing your branch on top of latest changes

If you have a branch that was previously created from master you may want to bring the latest changes from master into your branch. If you performed a simple git pull you will end up with merge commits in your history. To avoid the merge commits, the recommended workflow is to instead rebase your local commits in your branch on top of master.

Conceptually, the rebase operation will remove your current commits, bring your branch up to date with master, and then replay your commits one at a time on top of the latest changes. You will have the chance to resolve merge conflicts that may occur or safely cancel the rebase operation if desired.

To rebase your current branch on the upstream master branch run:

git rebase -p upstream/master

NOTICE: Be aware that performing the rebase will cause your local clone of a your branch and the fork version of your branch to diverge. In order to push the local branch to your fork you have to rewrite the history of the branch in your fork:

git push --force-with-lease

The --force-with-lease option will make sure that your fork's branch does not have any changes that are missing locally. This can happen if you perform development on a different machine, pushed them to your fork, and have not synced those changes locally.

DO NOT REWRITE HISTORY IN SHARED BRANCHES: Your fork is your playground for development and as long as you are the only developer contributing to a branch it is safe to rewrite the history of the branch. It is encouraged before submission to rebase your changes on top of master and to remove extraneous commits to improve the commit history. However, if multiple developers are doing development in a branch it requires coordination before rewriting history using rebase.

Creating hotfix releases

To create and release hotfixes, annotated git tags are used; no release branches are created and maintained. To create a hotfix release, follow the steps below:

Maintainer will create a hotfix branch from the desired tag in ni/VireoSDK
  • Go to the VireoSDK GitHub in your browser, click the "Branch:master" button and select the "Tags" tab in the pop-up that appears.
  • Choose the tag (say v10.1.7) on top of which you want to create a hotfix. The button should now say something like "Tag:v10.1.7".
  • Click on the "Tag:v10.1.7" button, select 'Branches', and type a new name for the branch (for example, 'hotfix'), hit Enter. This will create a new 'hotfix' branch in ni/VireoSDk from the chosen tag (v10.1.7).
Contributor will create a local branch with the desired fixes and do a pull request
  • Fetch new branches and tags from remote.

    git remote update -p
  • Create a local branch (say 'myHotfix') using the same tag as the 'hotfix' ni/VireoSDK branch.

    git checkout -b myHotfix v10.1.7
  • Then pick appropriate commits that should be cherry-picked into the hotfix release.

    git cherry-pick ec64f3296e5ee858dbe088768f0ff4fb4afad221
  • NOTICE: The selected commits will apply directly, 'git status' will not return pending changes.

  • Submit your branch to your remote repository.

  • Create a PR from your branch 'myHotfix' to 'hotfix' in ni/VireoSDK. Please comment in the PR if this is a straight merge.

Maintainer will merge the changes
  • Once your PR has been reviewed and approved, the Maintainer will merge your PR into the main hotfix release
Maintainer will push a new annotated tag to create the hotfix release
  • If this is the first hotfix for the tag, we need to switch to prerelease first.

    • Manually update the package.json file, append '-hotfix' to the version (10.1.7 becomes 10.1.7-hotfix). Save the file.
    • Run 'npm install' to update the package-lock.json
    • Commit just this change with just package.json and package-lock.json updates.
  • Once the package.json is in hotfix configuration, create new hotfix version with the following commands:

    npm version prerelease -m "%s Bug fix for ScanToString with booleans"
    git push --follow-tags
  • Once the tags are pushed, delete the 'hotfix' branch in ni/VireoSDK

Testing local Vireo changes

The Test.md describes how to test your Vireo changes locally on your development machine with the tests in VireoSDK. This section will instead focus on testing your local Vireo changes in other applications that consume Vireo.

Testing vireo.js in a JavaScript application

The vireo.js build is packaged as an npm package to make it consumeable in JavaScript applications. There are two primary workflows for testing your local Vireo changes in other npm based applications:

Testing vireo.js on your local machine

If you have a JavaScript application on your local machine that depends on Vireo you can do the following:

  • Checkout / make sure the package.json and package-lock.json are editable

  • In the directory with the package.json file run:

    npm install <PATH_TO_VIREO_SDK_DIRECTORY>

This will modify your package.json and package-lock.json to resolve the "vireo" dependency to a local path on disk instead of to a version number published in the npm repository. The VireoSDK directory will be symlinked to the node_modules folder of your application. This allows you to make changes to your local vireo and have them be instantly available to the consuming application.

WARNING Since the local VireoSDK is symlinked to the JavaScript application be aware of tooling that could accidently delete or modify your local development files unexpectedly.

Testing vireo.js on a different machine / CI

If you want to test your local vireo changes on a separate machine from your local machine (a CI or test machine) without publishing to the npm registry you can do the following:

  1. In your local VireoSDK directory run: npm pack
  2. This will create a vireo-<VERSION>.tgz file
  3. Move the tgz file to a location that can be bundled with your app. A good location would be next to the package.json and package-lock.json file
  4. With the package.json and package-lock.json files checked out / editable run:
npm install <PATH_TO_TGZ_FILE>

This will update the package.json and package-lock.json to point to the tgz file location. You can now share the application that has the tgz file, updated package.json, and package-lock.json to test.

Testing esh.exe in a .NET application

The esh.exe build is packaged as a nuget package to make it consumeable in .NET applications. There are two primary workflows for testing your local Vireo changes in other .NET based applications:

Testing esh.exe on your local machine

  1. If you do not already have nuget.exe, download it here. Version 4.7.1 should work.

  2. In the root VireoSDK directory file run:

    nuget pack VireoSDK.nuspec -properties version="[newversion]"

    replacing [newversion] with the version to update to.

    This command will generate a .nupkg file in the folder the command is run from.

  3. See installing a nuget package for the various ways to install the created package to your .NET application. If you are at NI and working on G Web Development Software, follow these steps:

    1. Run this command:

      .\BuildTools\RefactoringTools\UpgradeNuget.exe .\Source\VI\Tests\Tests.HtmlVI vireo oldversion newversion

      from the ASW root directory, replacing oldversion with the existing vireo version and newversion with the version you specified in the previous step.

    2. Add this line to the packageSources section of the NuGet.config in the root ASW directory:

      <add key="Local" value="Source\VI\Tests\Tests.HtmlVI" />
    3. Place the .nupkg file created in the previous step into Source\VI\Tests\Tests.HtmlVI.

    4. You should now be able to build and run .NET tests that will use your locally created Vireo.

Testing esh.exe on a different machine / CI

  1. Follow the same steps above for testing on your local machine.
  2. You can now create a change that includes the .nupkg file to test.

Configuring Visual Studio Code Intellisense for the JavaScript build

  1. Create a c_cpp_properties.json file for the VSCode C/C++ configuration. A quick shortcut for creating the c_cpp_properties.json file is to hit Ctrl+Shift+P from inside Visual Studio Code and search for and run the C/CPP: Edit Configurations... command.

  2. Add an Emscripten configuration to the c_cpp_properties.json file that uses the includePath (make sure to update the paths to point to your Emscripten installation directory) and defines shown in the following example:

    {
        "name": "Emscripten",
        "browse": {
            "path": [
                "${workspaceFolder}"
            ],
            "limitSymbolsToIncludedHeaders": true
        },
        "includePath": [
            "C:/dev/Github/emsdk/emscripten/1.37.36/system/include",
            "C:/dev/Github/emsdk/emscripten/1.37.36/system/include/libcxx",
            "C:/dev/Github/emsdk/emscripten/1.37.36/system/include/libc",
            "C:/dev/Github/emsdk/emscripten/1.37.36/system/lib/libc/musl/arch/emscripten",
            "${workspaceFolder}/source/include"
        ],
        "defines": [
            "kVireoOS_emscripten", "__EMSCRIPTEN__", "VIREO_DEBUG", "VIREO_USING_ASSERTS", "VIREO_DEBUGPOINT"
        ],
        "cStandard": "c11",
        "cppStandard": "c++14",
        "intelliSenseMode": "clang-x64"
    }
  3. After the configuration is saved, when you have a C or C++ filetype open you should see an environment configuration in the bottom right of the status bar. Click the environment configuration to choose the newly added Emscripten configuration.

Debugging toggles

Vireo has toggles that can be turned on to aid in debugging a problem. To turn on a toggle, look in the DebuggingToggles.h file for the toggle definition and change its value from 0 to 1 and then rebuild Vireo.

VIREO_DEBUG_EXEC_PRINT_INSTRS

Turn on this toggle to print the name of the actual Vireo instructions being executed

The following VIA program when run

define (CheckEqual dv(.VirtualInstrument (
   Locals: c(   // Data Space
       ce(dv(.Int32 -56)c1)
       ce(dv(.Int32 -56)c3)
       e(.Boolean local5)
   )
   clump(1
     IsEQ(c1 c3 local5)
   )
)))
enqueue (CheckEqual)

Will produce output similar to this in the console:

Exec: IsEQInt32
Exec: Done

VIREO_DEBUG_PARSING_PRINT_OVERLOADS

Turn on this toggle to print the overloads available for an instruction and the overload being selected as the instruction is being parsed. This is helpful when debugging why Vireo is not properly parsing a new Vireo instruction.

The following VIA program when run

define (CheckEqual dv(.VirtualInstrument (
   Locals: c(   // Data Space
       ce(dv(.Int32 -56)c1)
       ce(dv(.Int32 -56)c3)
       e(.Boolean local5)
   )
   clump(1
     IsEQ(c1 c3 local5)
   )
)))
enqueue (CheckEqual)

Will produce output similar to this in the console:

=========================================================
Finding an appropriate overload for 'IsEQ'
It currently has the following overloads:
        IsEQBoolean (Boolean, Boolean, Boolean)
        IsEQRefnum (EventRegRefNum, EventRegRefNum, Boolean)
        IsEQRefnum (UserEventRefNum, UserEventRefNum, Boolean)
        IsEQRefnum (ControlRefNum, ControlRefNum, Boolean)
        IsEQRefnum (JavaScriptStaticRefNum, JavaScriptStaticRefNum, Boolean)
        IsEQRefnum (JavaScriptDynamicRefNum, JavaScriptDynamicRefNum, Boolean)
        GenericBinOp (*, *, *)
                Generic loader
        IsEQRefnum (QueueRefNum, QueueRefNum, Boolean)
        IsEQUtf8Char (Utf8Char, Utf8Char, Boolean)
        IsEQDouble (Double, Double, Boolean)
        IsEQSingle (Single, Single, Boolean)
        IsEQInt64 (Int64, Int64, Boolean)
        IsEQInt32 (Int32, Int32, Boolean)
        IsEQInt16 (Int16, Int16, Boolean)
        IsEQInt8 (Int8, Int8, Boolean)
        IsEQUInt64 (UInt64, UInt64, Boolean)
        IsEQUInt32 (UInt32, UInt32, Boolean)
        IsEQUInt16 (UInt16, UInt16, Boolean)
        IsEQUInt8 (UInt8, UInt8, Boolean)

        trying... IsEQBoolean (Boolean, Boolean, Boolean)
        trying... IsEQRefnum (EventRegRefNum, EventRegRefNum, Boolean)
        trying... IsEQRefnum (UserEventRefNum, UserEventRefNum, Boolean)
        trying... IsEQRefnum (ControlRefNum, ControlRefNum, Boolean)
        trying... IsEQRefnum (JavaScriptStaticRefNum, JavaScriptStaticRefNum, Boolean)
        trying... IsEQRefnum (JavaScriptDynamicRefNum, JavaScriptDynamicRefNum, Boolean)
        trying... IsEQRefnum (QueueRefNum, QueueRefNum, Boolean)
        trying... IsEQUtf8Char (Utf8Char, Utf8Char, Boolean)
        trying... IsEQDouble (Double, Double, Boolean)
        trying... IsEQSingle (Single, Single, Boolean)
        trying... IsEQInt64 (Int64, Int64, Boolean)
        trying... IsEQInt32 (Int32, Int32, Boolean)
        An overload was found.
=========================================================
Finding an appropriate overload for 'Done'
It currently has the following overloads:
        Done ()

        trying... Done ()