WE are installing Azure Machine Learning SDK in Python Code Recipe in DataIKU.
The Model coding is more manual when using SDK, Is there a way where we can create models in ML Studio (Drag and Drop) and use it's service in Python to get the output.
This is a forthcoming feature of the Pipelines Designer UI, in which you can use custom Modules in the drag-and-drop fashion you propose. This UI will be built on top of Pipeline Modules which already exist today. Check out this notebook for how to create a custom module that is easy to plug-and-play using the SDK.
Related
I have a Xamarin.Forms application which targets iOS, Android and UWP. I have built an Azure pipeline that generates my .ipa and .apk files which can be used to install on devices that support its debugging. However for UWP I am trying to generate a build that will allow side installation (not for the app store), but not sure how to do this.
In visual studio there is the option to "create app package" (see screenshot below) which generates a bunch of files including an install.ps1 script. And running this script installs the application on my machine. I would like to automate this process in my Azure Devops pipeline but not sure how to. Any help would be appreciated.
Figured out how to do this.
I used the following MSBuild Arguments in my Build solution Task:
/p:AppxBundlePlatforms="x86|x64|ARM"
/p:AppxPackageDir="$(System.DefaultWorkingDirectory)\AppxPackages\\"
/p:AppxBundle=Always
/p:UapAppxPackageBuildMode=SideloadOnly
/p:AppxPackageSigningEnabled=true
/p:PackageCertificateThumbprint="$(UWPThumbPrint)"
/p:PackageCertificateKeyFile="$(UWPCertificate.secureFilePath)"
/p:PackageCertificatePassword="$(UWPCertificatePassword)"
$(System.DefaultWorkingDirectory) may be replaced with $(Build.ArtifactStagingDirectory) if you are using an Azure machine. See this link for information on Build Arguments.
I am then able to archive the contents of $(System.DefaultWorkingDirectory)/AppxPackages with an Archive task.
I have a .Net Core 3.1 application that I'd like to deploy as a Self Contained (SCD) Deployment using an MSI Installer in Visual Studio 2019...
I have published the project as an SCD Deployment and I can run this on the target machine just fine
However, I've been asked by IT support to provide an MSI installer for the application as .MSI files work well with some of the admin/control applications they have.
So I created a setup project in my solution and selected 'Publish Items' as the Project Output of the setup project and rebuilt it...
The installer seems to run just fine on the target machine but when we try to run the installed application it says that .Net Core is required ...It's as if the installer has ignored the Self Contained aspect and just installed as a regular Framework Dependent Deployment
Is there a way to create an installer that installs an SCD deployment? Have I made a mistake in my thinking?
Many Thanks in advance,
Andy
The VS Setup Project template is quite old. Most likely it was not updated to be "aware" of the SCD support so it resorts to extracting the classic output binaries from your project.
There are other free tools that you can use to create an MSI from VS, which give you more options to customize and correctly configure the package.
If you have time and want to learn a new skill, try WiX Toolset. It is very powerful but you will need some time to get started.
If you wanted to get it done quickly and avoid the hassle, use the free VS Extension from Advanced Installer. Its GUI allows you to easily create your setup package and it has native support for .NET Core packaging too. If you follow the steps from the linked tutorial it should create a working package for your application.
Disclaimer: I work on the team building Advanced Installer.
I have a medium sized C/C++ Qt application that uses some win32 API functions. I compiled for release and created a directory that contains all the needed DLLs and resources files needed. I tested it on other machines and it works fine.
Now my question is.. how to package such application so that I can put it on microsoft store? Note that it is not a UWP app and it uses some functions from win32.
You can use the Professional edition of Advanced Installer (it has Visual Studio integration) and create an MSIX package ready for the Windows Store.
Here is an article that takes you through the MS Store publishing steps
Also, an article that explains how to build and MSIX package using Advanced Installer's VS extension
Disclaimer. I work on the team building Advanced Installer.
So I've been using Azure ML Studio for a while and now Microsoft have come up with the new tool called Azure ML Workbench. The workbench seems to be pretty low level and it seems that the majority of functions need to be hand coded in Python.
So if I have an experiment in Azure ML Studio using some of the drag and drop Training methods such as Matchbox and Boosted Decision Trees. How can I convert these to run in Azure ML Workbench ?
You can't use ML Studio modules in Workbench, as these are separate tools with separate purposes.
Keep in mind that you don't necessarily need to hand-code the algorithms in Workbench - you're free to use libraries such as scikit-learn to get the job done.
Alternatively, you can deploy your ML Studio model as a predictive web service and use it as such from Workbench, but I wouldn't really recommend this approach (and it doesn't sound to me that this is what you'd want, either).
Is it possible / planned to build a Win/Mac/Linux package from within one platform?
An IDE called RunRev LiveCode allows to deploy to multiple platforms, but it supports only HyperCard language, afaik it's also possible in Qt
a Zotero Standalone Builder can be used to bundle Webapp XUL Wrapper into distributable bundles for Mac, Windows, and Linux
would it be possible to use it with TideSDK?
a Kickstarter project starts for AppJS, the maintainers want to launch a cloud service to deploy the app for all platforms in the cloud and then just download the exe, dmg or a linux package
It is not possible to build apps on the same platform with TideSDK. A service platform is coming that will solve this issue to make development easier for everyone. We'll make announcements with this available.