Check and Notify non-existence of Microsoft.VisualBasic.PowerPacks - vb.net

In a simple windows form application on VS 2010 I have put a ovalShape using power packs.
The simple Form
Now automatically this action puts the reference of Microsoft.VisualBasic.PowerPacks.Vs in to project properties.
when deploying this in different PC obviously the (a)powerpacks needed to be installed if this application doesn't work, (b) or it can set to "copy local = true" in reference properties so that it should sit to next with the application.
assuming (b) is not an option, since it is a solitary executable, (a) is the only option. in this way if the target machine does not have powerpacks the requirement is to notify it to the user in the first place.
apparently the dll will be deployed in when using the "VisualBasicPowerPacksSetup"
C:\Windows\assembly\GAC_MSIL\Microsoft.VisualBasic.PowerPacks.Vs\10.0.0.0__b03f5f7f11d50a3a\Microsoft.VisualBasic.PowerPacks.Vs.dll
so the blind approach is just to check if the above file not exist then prompt user to install "VisualBasicPowerPacksSetup". but i feel its more accurate if the application able to actually check in registry level.
in registry "Microsoft.VisualBasic.PowerPacks" records in several location, thus makes a confusion.
how to identify the correct key and what should be correct way of checking this reference in vb ?

You could just try to create an object defined in the dependency and catch the resulting exception.
Handling this you could ask the user to install the package. This is probably not considered good practice but should get the job done.

Related

Debug .NET com dll when running from Access

I have a com dll developed in VB.net that provides an interface to eBay api's. Been working for many years. Recently eBay change certificate authority and moved to TLS 1.2. Sorted through that but now the dll seems to quit in mid transaction when called from Access. The functionality does listings from local inventory to eBay. The first part loads pictures of the listing item to eBay. That works fine. I can see the calls and responses using Fiddler. It then moves on to actually listing the item which seems to stop when called from Access. What I get back in Access is "Can't find x.dll at file location". I know the call is going to the dll because it uploads the pictures.
I have a test project developed in VB.net to test the dll and when run from that everything works. I can see the listing call and response in fiddler. When run from Access there is no listing call. I can verify that the endpoints for the calls are the same from test project or Access.
My question is how to debug the dll in the Visual Studio IDE when it is called from Access. Any thoughts?
Yes, the way you do this is open up the vs class you have for the .dll.
Then in debug of the project, setup this:
Now, when you hit f5, then access will launch and run.
If at this point, say you close access (exit), then you note the debugging process stops.
However, go though the forms and whatever in Access, and get to the point where you hit that button or whatever.
You find now that you can say set a break point and even step and debug your .net code like any other code. So say in my example (a custom sage 300 .net interface to access), I want to debug "find customer" routine from VBA.
Well, in vb.net I have this code:
so, when any routine - even those in the class instances is called, you can set break-points etc.
And to stop? Just exit access.
So, you can make a change to your code, and even hit f5.
It not clear how you are registering your .dll (or do you use some VBA to side-load the .net - that's what I do, since it is a pain to have to register my .net com objects on each PC.
But, do give the above a try. It should let you debug your .dll code. The trick is to have VS launch ms-access as a attached debug session - and that is exactly what the above setting in your class project above allows you to do.
do note in above, I used the path to access 2010 (access 14), so replace the path name to the access.exe with your version of access you are using.
And keep in mind, that if have some installer, or some custom "thing" that registers your .net .dll for you?
Well, during this process, if your VBA assumes a registered com object, then ensure you have this check box enabled:
On compile for the project settings, you thus want this option checked:
NOTE very careful, checking the above option does not change the code or anything at all - it ONLY does a regasum automatic for you, and this would of course re-register your existing .dll - which is what we want for debugging. After you done, do run your re-register of your .dll to switch back from the debug .dll that going to be in your current project bin folder to whatever you "regular" use on your PC.
FYI:
In MOST cases, I find this whole idea does NOT work unless you launch VS as administraor. So, make sure VS is being run as administraor for this to work.
I tend to just tap windows key, or even right click on your vs shortcut, and of course choose run as administraor.
This is so often required, you note that VS will EASY and quite CLEAR show you running in admin mode, and I quite much now always run VS this way.
You should see this:

Is there a way to set a property value to the formatted install date/time

Is there a way to set a property value to the formatted install date/time?
I'm in the process of creating an MSI installer for an old VB6 application we still depend on (yes, I know, upgrade before it dies).
I'm trying to add a backup folder for the user data files in the install-folder (not my application design, nor my application). Unfortunately every user of this application has their own copy of the data file installed on their system (dedicated machines per user), and the installer has the default file. I would like to create a backup folder so that I can manually (if necessary) go back and retrieve previous versions of the file.
What I'm thinking is
c:\program files (x86)\app*.mdb => c:\program files (x86)\app\backups\201804091125
This will be rushed. Please tell me what is not clear.
Custom Action: In order to implement exactly what you describe, you generally need a custom action. This is always unfortunate since they are very error prone: Why is it a good idea to limit the use of custom actions in my WiX / MSI setups?
Alternative?: If you ask me I would install the database in a component of its own, make the file the key file and set the component to permanent and never overwrite if key path exists.
In the WiX source: for the WiX component element, set these attributes: Permanent="yes" NeverOverwrite="yes"). I am not 100% sure what will happen if you do something stupid such as setting REINSTALLMODE="amus" during installation (force overwrite all files regardless of version). It has been a while since I tested the NeverOverwrite flag. But for normal deployment done the regular way the database file should be left alone and not overwritten.
Custom Action Overview: There are properties called Time and Date that are automatically set in the installer, but the Date property will generally contain characters that are illegal in path names. It is possible to just get the properties and replace the illegal characters. However, the date separation characters are probably different based on regional settings and hence hard to predict. Your code could get messy quickly and testing would be challenging (potentially many locales to test depending on distribution scope - a truly globally capable package is challenging).
I would rather get the date and time some other way - via some programming API call where I can determine what format the data comes back in. You also need to run this custom action elevated in deferred mode to ensure it doesn't fall over with access denied (insufficient user rights for operation). This is always quite a bit of clunk to set up and get working. Maybe try the alternative approach first?
I have long considered adding a custom action to abort the install if REINSTALLMODE="amus" has been specified. I would prefer that and the alternative approach described with "never overwrite" to a custom action doing all this copying.

IBM Urbancode Deploy - Supply values to parameters at runtime using properties

I have created a process in IBM UCD to deploy a .Net application.
My Scenario is that i should be able to provide different application name at run time each time i run the process. How can we do this using property in IBM UCD.
I have tried enabling "Prompt on use" option and also created component property and mapped it to the parameter say ${p:component/application.name} but doesn't seem to work. May be i missing out some sequence of steps.
It would be great if i get detailed steps to making this working.
I take it that you are on version 4.x (uDeploy)?
I would steer clear of the prompt on use approach, that feature was removed in 6.x. While there is a migration in place, its simpler to just avoid it.
Using a property on the component process itself is the way to go. So go to your process configuration, and go to the properties / configuration tab. Create a property there. You'll be prompted for a value whenever you run an application process that uses this component process.
If the property is named "iis.app.name" you would reference it with just ${p:iss.app.name}.
Don't use the property "application.name". That is an automatically created property that gets the name of the UCD Application that you are deploying. If you ever can't find out the right way to reference a property, look at your executed process (at component / application levels). The normal view that lists out all the steps that were run and how long they took is sitting on a tab called "Log". Right next to it is "Properties" tab. Click that and you'll see what properties were available to the process.
Also, you'll have better luck getting fast answers about UC Deploy using their own forum: https://developer.ibm.com/answers/?community=urbancode
Did you tried using process plugin for updating the property file ?
Application >> Process >> Select Process >> Process Editor -- From left panel you can Utility plugins , try with update property option.

Changing the location of an existing VBA Reference from C:\Windows\system32\ to a shared drive

I've been trying for a while now to get a reference file to be loaded externally to no avail.
To be specific I am trying to load a "Microsoft Date and Time Picker Control 6.0(SP4)" which usually resides in C:\Windows\System32\MSCOMCT2.OCX
However some people that run a macro containing this element don't have that "MSCOMCT2.OCX" file on their PCs so I thought I will move the MSCOMCT2.OCX to a shared location and reference the code to use the shared one instead (so everyone will have access to it)
I tried doing that but when I was trying to load a reference with "Browse" from a different location it didn't load it - because I already had that in C:..
So I thought OK... I will remove the file from C:\ so I can only reference the shared file. - so I deleted it.
So I open the workbook again and look at references - I cant find "Microsotft Windows Common Controls-2.6.0(SP4)" - great!
And I proceed to add it manually with Browse from the shared drive.
When I do that however 2 references of "Microsotft Windows Common Controls-2.6.0(SP4)" are being added - 1 from C:\(which is not there) and 1 from the shared drive.
The one from C:\ is always automatically selected.
If I try to disable the one from C:\ and enable the one from the shared drive it automatically changes back to what it was when i press OK.
If I try to enable both - it says duplicate References and keeps only the one from C:\
So.. does anybody know how can I get rid of that C:\ reference from the list so it doesn't get loaded? Apparently deleting the files themselves did not work.
Ultimately my goal is to enable people without C:\Windows\System32\MSCOMCT2.OCX file to be able to use my Date Picker Tool.
Thanks a lot!
ActiveX control rereferences are always GUID-based. The VB IDE shows you the current location of the file as listed in the registry on your computer, as a courtesy, but it really doesn't matter what it says. The control will be loaded from wherever it was registered on the user's computer.
That's the key: the control must be registered on the user's computer.
I must strongly discourage you from doing what you're trying to do. You might be able to concoct a method by which you load the DLL from a network location, but it presents no advantage over doing the Right Thing(TM), and plenty of problems. The Right Thing is simply that if you need that control, you must distribute and register it with your application, just like everybody else does. And you really should install it in the recommended location for it (System32); not on the network.
Here's a quick example of what can go wrong: you provide your user with you app, and it works with the control on the network like you want it. Then the user installs another application that happens to need the same control. The app's installer sees that the control is already registered on the user's computer, so it doesn't try to add it again. Except that this particular app is intended to be used when the user is not connected to a network. Now you just broke someone else's program.
The VB/VBA architecture was never intended to support XCOPY deployment. I'm know it's a pain and that these extra steps are extremely inconvenient when you're just trying to deploy a "macro". Sadly, it's the nature of the beast. I'm sorry

Class table generated InprocServer32 value problem

I'm installing an Active X control that contains some COM servers. I'm using InstallShield's COM Extract at Build option to generate the registry information. This results in a lot of entries in the Registry and Class tables. (The extracted information is pretty much the same using Wix).
It appears that my COM Sever is correctly being installed except for an additional value called "InprocServer32" in the InprocServer32 key that looks like this:
HKCR\CLSID\{MY-COM-GUID}\InprocServer32
(Default) = C:\Path-to-my\file.ocx
InprocServer32 = 8tYCAGak)9S9&~swl.$?MyFeatureName>*&N$B'fk?As1x2J653?'
The only think I can make out from the extra value is the MyFeatureName which is the internal name of the MSI feature that contains the .ocx file. The key is not listed in the Registry table so it must be generated by the Class table.
The problem I'm having only happens in Windows Server 2008. It seems that the app trying to use the COM server is failing to find the path to the .ocx file from the (Default) value and instead it is finding the InprocServer32 value. This results in the app launching the MSI and then having the MSI being stuck in what seems like an infinte loop.
I'm wondering if this is a known issue in Windows Server 2008 or whether there is a way to prevent that extra value from being generated by msiexec.
I'd read this article and see if it helps you get where you want to be:
RobMen's Recommendation: Do not advertise COM information in MSI
You might want to turn off InstallShield's COM Extract at Build and instead do a One-Time COM Extract on the component in question. Then you can go into the Component Advanced section and manually manipulate the registry / com table information to be how you want it to be.
If you use WiX at all, another workflow / trick is to use Heat to build an MSI or MSM around your COM server. Then use InstallShield to edit the MSI/MSM in direct mode and the Registry view to export the Registry Keys/Values to a .REG file. Then import that .REG file into your Component in your real install project.
I can't help you diagnose what's going on, I'll just mumble a bit about what this all means. This is a counter-measure against DLL Hell. It is supposed to protect your application against some kind of other install program that could overwrite your COM server registry keys. Specifically the (Default) key which gives the location to your server DLL.
From the fake InprocServer32 value, the app can auto-detect that the Default key was overwritten and automatically launch MSI again to repair the damage. Which is what you see happening.
I thoroughly dislike the feature, it is just one more fail point in something that is already hard to troubleshoot when it blows up. And it is a useless feature, it assumes that the other installer doesn't use the exact same counter-measure. Which would have worked 10 years ago.
No idea what you'd do to troubleshoot this particular failure. Other then just punt this cr*p and let the servers just SelfReg themselves. At least you'll have something to work with when that doesn't work.