I am am considering installing a dll assembly in the GAC and I don't have enough experience to know if this is the best practice for this scenario. Hopefully someone here can shed some light on me and get me pointed in the right direction.
The Scenario:
I have a library assembly that I have written. This DLL is designed to read some data from an output file of a third party software package. Once read the data is available for access to the caller of this DLL.
My company has developed four different windows form applications that use this library file to get data into the apps.
The 3rd party software manufacture has change the the way they write their output file, so I have modified my DLL to support the new features the third party is using.
Currently all the apps have been written using the dll as a private assembly.
The Update:
To update all of the applications would require a rebuild of all four apps referencing the updated dll file, then reinstall all four apps on every machine that may be running them.
If I were to change the dll to a GAC assembly, then all I would have to do is reinstall the updated dll in the GAC assemble on every machine that uses the DLL assembly.
The later option seems to be simpler, then the next time this happens, we just update the one dll in the GAC and away we go.
The Question
What is the best practice here?
This link makes me think I should not be thinking about GAC
When should I deploy my assemblies into the GAC?
And if GAC is the answer how do you go about deploying the strong named assembly so that is installed in the GAC from the setup file?
Any help is appreciated,
Mike
Related
We are developing audio plugins which are dynamic libraries. This made it very hard to use 3rd-party libraries such as Qt, as our product DLL would search for Qt's DLL on host DAW program's executable directory, and it is impossible to install our dependent DLLs to that position (you don't know which host will be used, and host may even be installed later than the plugin).
I've made some brief search on this problem, the few answers direct me to a Microsoft technology called side-by-side assembly, and I'm almost drawn in the huge documentation and concepts on that. So my question is:
How to make my DLL to load several specific dependent DLLs located at path_to_plugin/MyPlugin_dependents/xxx.dll?
Or is there any examples to side-by-side assembly that simply do this and is much simpler than the official example?
I'm not very experienced with deploying applications and i'm having an issue with my very small screen scrape application.
I use a project that handles all screen scrapes and i just call the functions from the dll(on dev machines at least). When i reference that dll from the directory, it works fine on my machine but installing and launching on another machine without the project crashes it immediately when it tries to access the class to instantiate the screen scrape object.
Perhaps my assumption of dll is incorrect but doesn't having the dll mean it incapsulates all the classes/references/etc in that project so it can be used elswhere without having to lug the whole project with it?
What could be going on with this dll?
The dll assembly encapsulates the project, but not its references. Anything that the project needs to reference must be available in the new environment as well, whether in the GAC (global assembly cache), the local directory, in a reference path, or wherever.
If you are using Visual Studio 2010 or earlier, you can use Setup Projects to nicely gather all necessary references and package them into an install package for you. There may still be complications that you will need to troubleshoot sometimes, but it simplifies your deployment effort. In Visual Studio 2012 and later, Setup Projects were dropped, but there are other options, like WIX and Install Shield. And that is a whole other topic with plenty of Q&A on this site.
This is driving me crazy. I have developed a .NET COM DLL that is used by a VB6 DLL wrapper in order to update and replace some legacy functions in an application.
I am now trying to remove the requirement to use regasm on client machines so have worked out how to do that on a test DLL which all works fine.
I branched the DLL just in case and added an app.manifest file. Everything else worked out fine and I got it all working. The manifest is embedded and Visual Studio 2012 generates a mydll.dll.manifest file in the release folder.
Then I went back to the original trunk and added an app.manifest file (no point in merging as there were no code changes). I copied the contents of the branch into the app.manifest file and built the release version. The manifest is embedded in the DLL but no mydll.dll.manifest file is generated.
I know that it's not strictly necessary to have the mydll.dll.manifest file but I'd like things to be consistent (and for some reason the test process doesn't produce the same results with the trunk version) so how can I force it to be created?
This is a VB.NET DLL project so it doesn't have (or I can't find) the 'Generate Manifest' property drop down mentioned in the first answer here. How can I set this? Or is there a way to set it by editing the project file directly?
References:
Original walkthrough article and some corrections.
Overview by Junfeng Zhang in two articles plus a useful tool
You are making a fairly common mistake. A reg-free COM manifest helps an application find a COM server without looking in the registry to locate the DLL. Embedding the manifest in the DLL is like trying to solve the chicken and egg problem, Windows cannot possibly find that manifest if it cannot locate the DLL first.
The manifest needs to be part of the client app. Which is tricky since it is VB6, it doesn't support embedding manifests in its executables.
You could tinker with the mt.exe tool, an SDK utility that supports embedding manifests in an executable. You'd have to run it by hand after building the VB6 binaries. That's unfun and very likely to cause trouble when you forget. It is in general not a joyful tool to use, documentation is meager, incomplete and unhelpful, a chronic problem with manifests.
The fall back is a separate app.exe.manifest file, what Windows will look for next when it cannot find a manifest embedded in the executable. Where "app.exe" must be renamed to the name of the VB6 program. The EXE, not the DLL. This now also gives you a chance to avoid having to register the VB6 DLL, presumably what you really want if you truly want to make your program run reg-free. The disadvantage is that it will not work when you debug your VB6 program, wrong EXE. You'd also need a vb6.exe.manifest, located in the VB6 install directory.
Needless to say perhaps, very hard to get ahead with VB6 here. It just wasn't made to help you do this, they didn't have a time machine in 1998.
I have to admit that I don't know VB at all, but in the case of C++ and C# Visual Studio projects I previously had to resort to calling mt.exe in a post-build step in order to get the DLL manifest I wanted. Maybe that workaround would work in your case as well?
I have an environment where there is a combination of managed and non-managed executables being run for a particular suite.
I have a scenario where the only reason I have a managed assembly (in this case) is because code was easier to write in C# than in VB6.
My 'main' application, written in VB6, needs to make a call to this assembly. So I create the TLB file, so its COM callable.
Now for deployment, is it better practice to place the managed assembly 'with' the tlb in the windows system directory? to have them both in the same place? Seems to me a bit easier to manage that way.
Or the managed assembly (not-signed since it isn't to be shared among diff apps) should just go to the BIN of the application dir, and the TLB moved to the windows\system32 and registered there.
Overall it seems more straight forward to keep them together, but it doesn't seem to be the correct way to do things.
Thx for any tips.
None of the above. You only need the type library when you compile your VB6 program. It does not need to and should not be deployed on the user's machine. There's a small corner case where a type library can help with marshaling an interface across an apartment or process. That won't be the case in a VB6 app.
Personally, I'd put both of them in your application's installation folder or under the Common Files structure.
The System32 folders are ideally just for the system files. (Ignoring history for the moment :)
In our environment, we have two in-house frameworks and a separate website. During development, the references to the in-house frameworks tend to be set tp project references. However, once we move to release, the in-house frameworks are installed into the GAC as they are used for multiple instances of the website on each server. All of the ProjectReferences are changed, by hand, to References and the website assemblies and website are re-compiled and deployed.
I am attempting to automate this process. What is the best way to handle these issues? I have started to learn MsBuild in the attempt to accomplish this, but am totally confused. Any pointers and/or suggestions on how to proceed?
As far as I am aware there shouldn't be any need to change from using project references to file references simply because the assembly is registered in the GAC - as long as the referenced project is strongly typed at the point that it is built, the reference should still be valid.
Clarification:
There is no fundamental difference between a GAC reference and a file reference - if you have a reference to a strongly named assembly, and that assembly is placed in the GAC then the end application will load that assembly from the GAC.
See this link for more information on how the runtime locates referenced assemblies:
http://msdn.microsoft.com/en-us/library/yx7xezcf(VS.71).aspx
FYI - I believe it is recommended that you don't reference assemblies which are in physically in the GAC, and that you instead reference the strongly named assembly before it is placed in the GAC (not to be confused with whether or not the assembly should be installed to the GAC on the end-user machine)
If you leave the project references in, MSBUILD will copy all the project references as DLLs into the target website's bin directory, even though you've installed these DLLs into the GAC, which I assume is why you want to change this at the release level--so that space can be saved.
I also assume that you are intending to do this at the Release Branch only, and not at the development branch of your source control. Because if you are doing this at all branches of source control, then doing by hand is actually the best and fastest way of doing this for a one-time change.
To do this in MSBUILD, you will need to create a custom task that can modify the project file. The project file is an XML file, so you could use XPath as your custom task in MSBUILD. In the project file, you will find tags for "" to your referenced projects. You will need to change those to "" tags instead. Use existing examples of the tag in your project to see how it should look
Sorry I don't have sample code for this--doing it this way will be a bit of an undertaking, as there will be a number of things to take into consideration, and it might end up not being feasible.