Choosing which runtimes to support - .net-core

I'm working on a .net core library/application (it's more of a plugin of sorts) that will run on windows x64 and a few flavours of linux. I have a few very large native dlls that will need to also be included in the release and I'd like to choose wisely which to include (windows x64, ubuntu and osx only). This is how bin/Release/net6.0/runtimes looks like:
I'd like to only distribute a couple of these (for size reasons), but if I delete any one them the application fails to start, even if they are not relevant at all. There seems to be some dependency check. Is there any way to tell it to either not do that check and trust me that the needed files are there, or let me select which ones to include? I know there is a RuntimeIdentifiers property but it doesn't seem to limit the distributions in any way
NOTE: For reasons I don't want to get into I can't make separate builds for each OS. It has to be a single deploy that works on all targeted systems.

Related

Porting a library to .NetCore. How do I know which nuget references are 'full' .NetCore (and therefore Linux/Mac compatible)

I am currently tasked with porting an code library over to .NetCore. Fundamentally, this is going fairly smoothly but there is one thing that is causing me concern and that is retaining platform independence.
Core is designed to run on Mac and Linux so I want to ensure that the library will also run on Mac and Linux when I'm done. However, in order to get things working I find myself including a lot of non Microsoft.AspNetCore.* nuget packages. (e.g. System.Diagnostics.Process, System.Net.Http, System.Threading.Thread, etc)
Clearly this won't be a problem on Windows but I do wonder if it would cause problems if say, someone targeting Linux was to include a reference to my library?
If it's not a problem then great but if it is then how do I know which nuget packages will be okay in a multi-platform library? (e.g. Is it only the AspNetCore ones that will work cross-platform?)
The System.* packages are (usually) from https://github.com/dotnet/corefx, and the 99.5% use cases are going to work on macOS and (supported) Linux (distributions) the same as they do on Windows.
There are cases where the System.* packages don't work the way you'd expect them to. Mainly this comes from times that .NET was adding convenience APIs to wrap Windows behavior. So reading a registry key, for example, won't work (and even in a future where the registry API works, it certainly won't be backed by a legitimate hive, so reading obscure values will report key not found). Mostly when an API isn't going to work on Linux/macOS it'll throw a PlatformNotSupportedException, instead of producing some sort of unexpected/wrong result.
So, as long as what you're doing doesn't feel specifically "Windows-y", everything from a netstandard* or netcoreapp* RID should Just Work. And if it doesn't, you can always file an issue at the corefx github project.

Configuring CMake to build with Xcode

I am involved in development of a large cross platform project that build for Windows, Linux, and Mac OS X. The build for the software is configured with CMake.
The CMake scripts have been designed to configure successfully for Visual Studio on Windows, and Makefiles are currently used for building on Linux and Mac OS X.
Pretty much all of the development for the project so far has been done with people working on Windows, and a little bit of work on Linux. I am interested in developing for the project using Xcode 4.6 on a Macintosh running Mac OS X 10.7, and I have encountering problems as the CMake files do not seem to configure properly for that development environment.
For non-windows platforms many custom commands have been written to try to configure things such as copying needed files or setting environments that are needed for certain operations such as running unit tests during the build process.
It seems that because Xcode is an integrated development environment simliar to Visual Studio is has this concept of a build configuration, and when software gets build output files in up in a directory path that includes that configuration concept (i.e. many build files end up in a path that ends with folder named something like Debug, Release, etc.)
CMake is supposed to have support for dealing with this build configuration concept and the mechanism utilized work well for Visual Studio. That do no seem to work for Xcode. For example our build engineers have design CMake scripts so that for Windows, many path and whatnot are configured using the CMAKE_CFG_INTDIR value which helps to qualify the build configuration.
The use of CMAKE_CFG_INTDIR is not working for Xcode as the script for Macintosh were written with Makefiles in mind which don't really have the build configuration concept. The use of CMAKE_CFG_INTDIR within custom commands used to configure things fails on the Macintosh as the value resolves to $(CONFIGURATION)$(EFFECTIVE_PLATFORM_NAME). This values are not define when the custom commands are run, so values are not set properly and build operations fail.
It is unclear what is needed so that the system can successfully configure for Xcode. Searching on the Internet so far has not yielded insight into what should be used to make sure that build configuration can be successful. What resources are available that would help in figuring out how to configure this project to build with Xcode?
If you're talking about custom commands set using add_custom_command, then you should prefer "generator expressions" to avoid issues regarding per-configuration build directories. From the docs for add_custom_command:
Arguments to COMMAND may use "generator expressions" with the syntax "$<...>". Generator expressions are evaluated during build system generation to produce information specific to each build configuration.
For example, the build directory for a target called "MyExe" could be referred to as $<TARGET_FILE_DIR:MyExe>
Generator expressions are available in a few CMake commands, not just add_custom_command.
If you have more specific problems, it's maybe worth asking further question(s) with the relevant details.

CSS cross browser compatibility on Ubuntu

I'm currently working in web development and my default desktop is Ubuntu and I'm kind of happy with the setup and applications I got going. But I need to test web pages for cross browser compatibility while still being on Ubuntu.
I have gone through hell trying to get IE7 or IE8 (with wine) to run on ubuntu and when they finally worked they were very buggy and the graphics/scrolling was insanely slow.
Of course there is the option of virtual box but again, too much GBytes just to run a small application!
So to all the CSS gurus out there, how can I continue with my beloved Ubuntu and still deliver a good quality (tested) page.
Thank you.
Edit:
Update for freshness:
I now use the paid service from browserstack.com to provide the multitude of different browser testing environments via flash tunnelling. I'm a paid user, but there is an initial free trial period. browserstack has freed me of the need to run the windows os on my machines in any form, virtual image or otherwise. Since it also allows tunnelling, I can host the site on my local machine but still test in browserstack browsers. I consider the monthly fee money very well spent.
End Edit
Various options I have tried, including "the final solution": free downloadable windows testing OSes from microsoft
I've tried a number of the options below, but virtualbox may be your best bet for full & complete testing, especially because in a professional capacity you often have to test ie8, ie7 -and- ie6. Which gets tricky with only a single os installed. So in order of simplest and most shallowly testing to most complex and most fully testing:
browserlab.adobe.com
A newer, interesting online solution is: browserlab.adobe.com. It's actually very specific and fast compared to browsershots. It only gives you screenshots, but it's a great first step. So I do recommend that for purely visual (and thus relatively shallow) testing.
Browsershots.org
And while browsershots.org is also something that you should use for an overview experience of what users might see, you really can't get by without the real browsers for javascript and behavior testing (instead of just display & rendering testing that browsershots provides). The delay before you can see the images is also killer.
Dual booting into windows
Another that I've tried is dual booting, I work 99% of my time in ubuntu, and I have windows installed & available to dual boot into. Not a fast way to test, but if you don't have any other way to access ie, it should work for at least the latest version.
Remote desktop-ing over to a running windows box
Before I mention the "covers-all-the-bases" option, another useful possibility is to set up a windows machine and boot it up and connect to it via remote desktop so that you can work from one machine and test from both.
The final solution, using virtualbox
Finally, the mother of all solutions, using virtualbox:
Luckily (I know you said you didn't like the virtualbox solution, and I know it's an annoying setup process, but...) Microsoft provides available-for-a-year-or-more virtualmachine distros with different versions of ie pre-installed, available without the need for a license for a year or so before you'd have to update the virtualmachine, #
http://www.microsoft.com/downloads/details.aspx?FamilyId=21EABB90-958F-4B64-B5F1-73D0A413C8EF&displaylang=en
Installing a virtualmachine from microsoft's freely available browser testing images
Because this guide to setup on ubuntu is no longer available in full anywhere else, just in case you or someone else actually need it I feel compelled to include the actual details of the install process that were suggested to me on the ubuntu forums and worked when I went through them. I apologize for their length. Courtesy of the now anonymous original poster on the ubuntu forums:
Free Access to Microsoft Browser Compatibility Virtual OSes, Install Steps for Ubuntu
http://ubuntuforums.org/showthread.php?t=1097080 (Ed: I can't find this thread online any more)
HOWTO: run IE6, IE7, IE8 on Linux in
VirtualBox You need: virtualbox, qemu,
wine
Code: apt-get install virtualbox qemu
wine
Download the free(!) Microsoft
Internet Explorer Application
Compatibility Check VPC Images here:
http://www.microsoft.com/downloads/details.aspx?FamilyId=21EABB90-958F-4B64-B5F1-73D0A413C8EF&displaylang=en
(Note: you don't have to download the
full pack, you can cherry pick
specific combinations of XP/Vista and
IE6-8)
Extract the VPC image(s) with wine
(double-click). (Note: it might take a
while before the first window shows
up)
Turn the VPC image(s) into (a) VMWare
image(s) (which is/are readable by
VirtualBox): qemu-img convert -f vpc
image.vhd -O vmdk image.vmdk
Setup a new VM in VirtualBox, using
the vmdk image as an existing disk.
Boot it, you will see the Windows boot
progress bar and ... it will BSOD
shortly after.
Fixing the BSOD:
The BSOD is caused because the virtual
Windows tries to load processor
drivers for the wrong processor (it is
not running on VirtualPC proc, but on
VirtualBox proc). Or something like
that... We need to force Windows not
to attempt to load drivers for the
processor (it doesn't need any proc
drivers, because it's all virtual
anyway). Start safe mode by
(frantically) hitting F8 at Windows
boot and choosing safe mode.
Ignore all the 'New hardware' detected
warnings (we will deal with those
later). Start a command box and run
the following command to disable the
loading of processor drivers:
Code: sc config processor start=
disabled (note the space between '='
and 'disabled'!)
Restart the virtual Windows, it should
now boot all the way to the Windows
Desktop.
Now just when you think you can start
browsing the web with IE, you will
find out that the virtual Windows
needs to install the drivers for the
AMD PCnet NIC, which are located on
the Windows install disk. Fortunately
for those without a Windows install
disk, there is another way :)
Download AMD PCnet drivers here:
http://www.amd.com/us-en/ConnectivitySolutions/ProductInformation/0,,50_2330_6629_2452%5E2454%5E2486,00.html
Make an iso file containing the
drivers. I used Brasero for
simplicity. Choose to create a Data
Project, add the zip file (or the
unzipped files, saves you a step in
Windows), create the iso. No need to
burn an actual cd!
Stop the virtual Windows, edit the
settings in VirtualBox: mount your
brand new iso.
Start the virtual Windows, when it
asks to install the drivers for the
PCnet nick, point it to the (unzipped)
drivers. Et voila! You have teh
innernets! (Now you can also try to
install the other drivers it complains
for, but it's not really necessary)
The image README says the image will expire after about a year. In my experience the system gets hobbled against multi-hour use, but is still usable for the kind of short periods that you might want when booting up to test a website. At worst you might have to go through these steps again, so be sure to put them somewhere where you can find them again after a year or so.
I think setting up a virtual machine (Virtualbox or VMWare or...) with a proper Windows will be your only (local) option.
I you don't have one, buy a used Windows XP license. XP is cheap (around 20-30 euros here in Germany, for example) and all relevant versions of IE run on it. Home edition is enough. No need for Windows 7 or anything.
You could install IETester on that to get all the IE versions on one OS. IETester has flaws and is not always 100% reliable in what it renders, but for a general CSS compatibility check it should be okay.
I've never tried IE using Wine, but even trying to imagine the combination gives me goose bumps :D
If you have a copy of Windows you could install it in a virtual machine (Virtualbox is a good, free option). Or if you don't mind a lot of lag time and publicly exposing your web pages you could use a service like BrowserShots.
I have not tried this on Ubuntu or anything but windows - but this seems to be a pretty good testing system over the web.
http://spoon.net/browsers/
however, I think your best result would be to use a VM if possible.
I have to add my voice to those opting for VirtualBox.
VMs are the only way to get an accurate representation of how IE platforms behave. They also allow you to keep your main Linux install free of WINE and IE gunk, which is otherwise always troublesome and fragile. (Especially if you're trying to run multiple IEs, which is unreliable and inaccurate even under Windows).
They're not necessarily that big, if you take care to prune the unneeded features, turn off swap, compact the disc image and so on. My XPSP3 test image is just over 800MB.
I didn't want to install all this stuff as I wanted to move forward quick.
I found public AWS images with pre installed browser that you just can start and use.
http://www.hens-teeth.net/html/products/cross_browser_testing.php
If you already have an AWS account this will take you only 5 min. Make sure that you enable the RDP port on the incoming traffic in your security group.
As I use ubuntu I was looking for a way to connect from it to MS Win.
I'm connection on to them via remote desktop.
The way to go here is rdesktop, a command line utility for Windows Remote Desktop. (sudo apt-get install rdesktop)
If you feel like a GUI use tsclient. It's very close to the windows version.
From a work flow perspective I develop for Chrome in Ubuntu first, then have a look at the other browsers via browserlab.adobe.com.
After that I start my new AWS instance to debug.
The small AWS Windows instance is a $0.12 per hour (http://aws.amazon.com/ec2/#pricing). I can work for a long time on that before it's worth installing all this stuff.
CrossBrowserTesting.com works from Linux. Allows you to access Mac, Windows, and Ubuntu configurations and all the browsers loaded on them via vinagre vnc client.

How can I enable auto-updates in a Qt cross-platform application?

I love applications that are able to update themselves without any effort from the user (think: Sparkle framework for Mac). Is there any code/library I can leverage to do this in a Qt application, without having to worry about the OS details?
At least for Windows, Mac and user-owned Linux binaries.
I could integrate Sparkle on the Mac version, code something for the Linux case (only for a standalone, user-owned binary; I won't mess with distribution packaging, if my program is ever packaged), and find someone to help me on the Windows side, but that's horribly painful.
It is not a complete solution, but a cross-platform (Windows, Mac, Linux) tool for creating packages for auto-updates and installing them is available at https://github.com/mendeley/Update-Installer. This tool does not deal with publishing updates or downloading them.
This was written for use with a Qt-based application but to make the update installer small, standalone and easy to build, the installer uses only standard system libraries (C++ runtime, pthreads/libz/libbz2 on Linux/Mac, Win32 API on Windows, Cocoa on Mac, GTK with fallback on Linux). This simplifies delivering updates which include new versions of Qt and other non-system libraries that your application may depend on.
Before considering this though, I would suggest:
If you are only building for two platforms, consider using standard and well-tested auto-update frameworks for those platforms - eg. Sparkle on Mac, Google's Omaha on Windows or auto-update systems built into popular install frameworks (eg. InstallShield). I haven't tried BitRock.
On Mac, the Mac App Store may be a good option. See https://bugreports.qt.io/browse/QTBUG-16549 though.
On Linux, consider creating a .deb package and a simple repository to host it. Once users have a repository set up, the system-wide software update tools will take care of checking for and installing new releases. The steps for setting up a new repository however are too complex for many new Ubuntu/Debian users. What we did, and also what Dropbox and Google have done, is to create a .deb package which sets up the repository as part of the package installation.
A few other notes on creating an updater:
On Windows Vista/7, if the application is installed system-wide (eg. in C:\Program Files\$APPNAME) your users will see a scary UAC prompt when the updater tries to obtain permissions to write to the install directory. This can be avoided either by installing to a user-writable directory (I gather that this is what Google Chrome does) or by obtaining an Authenticode certificate and using it to sign the updater binary.
On Windows Vista/7, an application .exe or DLL cannot be deleted if in use, but the updater can move the existing .exe/DLL out of the way into a temporary directory and schedule it for deletion on the next reboot.
On Ubuntu, 3rd-party repositories are disabled after distribution updates. Google works around this by creating a cron-job to re-add the repository if necessary.
Shameless plug: Fervor, a simple multiplatform (Qt-based) application autoupdater inspired by Sparkle.
Shameless plug: this a relatively old question, but I thought that it may be useful to mention a library that I created recently, which I named "QSimpleUpdater". Aside from notifying you if there's a newer version, it allows you to download the change log in any format (such as HTML or RTF) and download the updates directly from your application using a dialog.
As you may expect from a Qt project, it works on any platform supported by Qt (tested on Windows, Mac & Linux).
Links:
Website
GitHub repository
Screenshot:
Though it works a bit differently than Sparkle, BitRock InstallBuilder contains an autoupdater written in Qt that can be used independently (disclaimer, I am the original BitRock developer). It is a commercial app, but we have free licenses for open source projects.
I've developed an auto-updater library which works beautifully on Mac OS X, Linux and pretty much every Unix that allows you to unlink a file while the file is still open. The reason being that I simply extracted the downloaded package on top of the existing application. Unfortunately, because I relied on this functionality, I ran into problems on Windows as Windows does not let you unlink an open file.
The only alternative I could find is to use MoveFileEx with the replace on reboot flag, but that is awful.
However, renaming the working directory of the application works on Windows 7 and Windows XP. I haven't tried Windows Vista yet.
I have found WebUpdate to be quite useful, though it's written with the wxWidgets. But don't worry, it's a separate app which handles your updates. The steps to integrate it are pretty simple - just write two XML files and run the updater. And yes, it's cross-platform.
The advantage of it is it will automatically download and unzip/install all you required and not just provide a popup with a notification about a new version and a link to download it. Another thing you can do with it is customizable actions.
Project's main page is here, you can read the docs or take a look at the official tutorial.
The blog post Mixing Cocoa and Qt may solve the problem for the Mac platform.
You can use UpdateNode which gives you all the possibilities to update your software. It's using a cross platform Qt client and is free for Open Source!
UPDATE
Just did some further analysis on that and really like this solution:
Pros:
Free for Open Source!!! Even the client is Open Source: https://github.com/updatenode/unclient
The client is already localized in several languages
Very flexible in terms of updates. You can even update single non-binaries.
Provides additionally a way to display messages though the client.
Ready to use binaries & installer for all common Linux distributions, single Windows binary, as well as installer and a solution for Mac (which I have not tried, as I don't have a Mac)
Easy to use web service, nice statistics and update check is integrated within few minutes
Cons:
I am missing a multi-user management in the online service. Maybe they will do it in future - I will definitely suggest that in their feedback portal
The client is a GUI client only - so, you will need to shrink it down to run without a GUI frontend (maybe only necessary for people like me ;-) )
So, bottom line, as this solution is quite new, I think there is lot of potential here. I will definitely use it in my project and I am looking forward for more from them! Thumbs up!
This is an old question but there is not Squirrel in answers which is BEST SOLUTION , here is what I'm doing in qt 5.12.4 with qt quick "my qml app" you can do this in any other language
I'm doing this in windows there is mac version of squirrel too, I don't know about Linux
download nuget package explorer release
https://github.com/NuGetPackageExplorer/NuGetPackageExplorer/releases
open nuget package explorer and add this directory 'lib/net45' it doesn't matter you have a .net app or not, I did this for my qt application otherwise it won't work.
add all files into this folder specify your version in the metadata
save nupkg file
download squirrel release https://github.com/Squirrel/Squirrel.Windows/releases
add squirrel to windows environment path
open cmd and cd to directory of nupkg file
squirrel --releasify file_name.nupkg -> now inide releases folder, there should be setup.exe file which will install app and other files.
to create new version do 2,3,4,7,8 again if its an update it will create delta file which is only needed file to update, put this files into your service directory for example in updates folder of your website which you need to disable directory browsing in IIS , and to auto-update application you need to call Update.exe which is in parent folder of application root directory appdir/../update.exe --update http://yourserver.com/upates/ after application restart app should start with new version
you can find documentation for squirrel in https://github.com/Squirrel/Squirrel.Windows/blob/develop/docs/getting-started/0-overview.md and nuget package explorer here https://github.com/NuGetPackageExplorer/NuGetPackageExplorer and you can use only nuget.exe too if you don't want to use nuget package explorer which can be used for dynamic generation of versions, which can be download from https://www.nuget.org/downloads
That easy. Now you have auto-update app which will download updates from the server and auto-update app. For more info you can read documentations.
note: for iis uses https://github.com/Squirrel/OldSquirrelForWindows/issues/205
I suggest you read on plugin and how to create and use them. If your application architecture is modular and be split into different plugins. Take a look at Google Auto Update utility http://code.google.com/p/omaha/. We use this.
Thibault Cuvelier is writing a tutorial (in French) to develop an updater. I know the explanations are in French (and everyone is not understanding French), but I think this can be readable with a web translator like Google Translate. With this you will have a cross-platform updater, but you need to write it by yourself.
For what I know, the only part of the updater that is explained in the tutorial, is the file downloading part. In the case this can help you, refer to the tutorial, Un updater avec Qt.
I hope that helps.
OK, so I guess I take it as a "no (cross-platform) way". It's too bad!
I have found a solution that can be automated with built-in self-extracting patches and updates. for windows. I have started using their sdk. take a look at the massive documentation here, https://agersoftware.com/docs/ the sdk is called securesdk and comes with their app, SecureDelta sdk. does a great job on any kind of files, better results than lzma-included delta updaters

Packaging to use to deploy cross-platform?

On windows applications are typically packaged as MSI, on Redhat Linux as RPM, what would be a best open source packaging method that could be used to deploy applications to all platforms including different flavors of unix and windows?
Contents would include exes, unix binaries, java jar files, user data, even database scripts to be run.
(I recognize contents would vary per destination OS, ie. binaries would be different, win exe vs unix binary etc, but for example config files may be the same or in the case of java even the bytecode jars)
Key feature I'd like the packaging to support is different users and permissions for different directories, however I recognize supporting this feature multiplatform may be very difficult.
Rather than build a package that is supposed to work across all of your platforms, which is likely impossible, you should have your build system build different packages for each target platform.
With CPack (It come with CMake) you can create packages for Windows (with NSIS), Linux (rpm and deb), and OS X with "make package". CMake also simplify cross-platform building.
For a sample you can look at avogadro's CMakeLists.txt and AvoCPack.cmake
I have a client that uses IzPack to create a single installer (it's Java-based) that installs their app on Windows, OS X and Linux.
http://izpack.org/
NSIS is an open-source solution which, as far as I know is able to build installers that run on Windows and UNIX-likes alike. However, for software deployment on Windows (especially in corporate environments) MSI is the way to go and NSIS is more of a headache.
So I wouldn't advise that you try to build a single package/installer for different platforms. But rather, as RibaldEddie indicated, multiple packages: one for each platform. That also allows to restrict the contents of the package to the files relevant to each platform.
If you'd like to support packaging for multiple distributions, I'd suggest helping the packagers for those distributions out; use some sort of well-known build system for your software (GNU's autotools or something like scons or waf), and document the build, optional dependencies, and so forth pretty well.
That way, when a Debian, Ubuntu, Red Hat, SuSE, whatever, packager comes along, they'll be able to create the package for you. You can optionally include packaging templates for one or more distributions in a separate VCS tree that is available, if you'd like.
If you are looking at packaging a closed-source/proprietary application for multiple systems, you'd probably do best to package up a .tar.gz file and document the installation process for it. You'll also want to make sure that the build process used doesn't embed any path information into the application, so that it can be run in /opt, /usr, or /usr/local, which are some popular choices for third-party add-on software.
BitRock InstallBuilder allows you to create installer packages for each one of the platforms you mentioned (as well as creating RPM, DEB, packages etc. from a single project file)

Resources