ICE32 Error from MSM - installation

I have 4 MSM's that are needed for one of my products and they are all producing ICE32 errors when included. This would normally not bother me cause they are external errors and I still get my MSI and it works perfectly. The problem arrises though, that I can not request help from Microsoft with ANY issues that arise from this particular setup project if there are ICE errors (like good ole 32). Interesting side note is that the MSM's causing the errors are Microsoft modules.
Does anyone have any idea how to get rid of these errors short of editing the MSM files directly?
I have tried using EnsureTable element to no avail, and it is only these four MSM's that are used so I can't even try placing another merge module before them.

Turns out I was just ensuring the wrong tables. In total I had to ensure the Class, PublishComponent, Condition, TypeLib, and Extension tables in order to take care of the errors given to me by the VC_User MSM's from Microsoft. Hope this can help someone else find the problem a little more quickly.

Related

How to Validate a Cognos report in debug mode ( element-by-element or step-by-step)

We recently migrated from Cognos 10.1.1 to Cognos 10.2.1.1 ( 10.2.1 plus Fix pack1) . Some of our existing reports fail validation now.
From the cogserver.log file , it looks like the BIBUS Process is Crashing on validating the report.
We are working with IBM tech support via a PMR .
Wanted to try if someone here knows if it is possible to Validate a report step by step so that I can get some information or some logs on what element in our report is exactly causing the issue? i.e. Is it possible to do the report validation in a debug mode somehow?
Oh, what a wonderful feature that would be, but to my knowledge nothing like that exists at all. You could try setting the logging on your dispatcher(s) to the maximum to see if you can get any more informative errors.
I would start by trying to view the tabular data for each query individually. If you can identify which query (or queries) are causing your problems, then you can just remove items from the query until it doesn't fail, at which point you should have a pretty good idea of what the source of the problem is.
If that doesn't work, I would just start ripping major chunks of the report out and seeing you can get it to run. For example, if you have a report with 4 charts, delete half of them and try your report. Revert back to the original report, and delete the other half. Get it to work, and then start removing stuff from the half that fails until you can narrow it down to your problem.
It's kind of slow, but these approaches have always worked for me.
On a side note, we're about to make the same upgrade, I'd be interested in hearing what you learn.
EDIT:
Oh, forgot. Make sure you disable DQM and test your reports that way, if you haven't.
Unfortunately, there's no way to debug step by step. Finally got the Core dumps for the crashes, sent them over to IBM Folks ; and they identified it as a known bug in 10.2.1.1. So now we are at 10.2.1.2 (applied Fix pack 2) which solves the issue.

Lightswitch global query breaking app

I'm trying out Lightswitch for the first time, and I'm finding some of the 'built-in' features aren't as easy to implement as the videos show. The most notable being global queries, because everyone I create is causing a specific error that I cannot find a cause for. I've stripped it down to a query that does nothing but return a 'select *' on the table, and removed any code I've added.
Here is the query in my solution explorer:
And here it is in the designer:
Simple enough, so I add a screen for the query:
Everything is pure default, so why do I get this error? It says it doesn't recognize the query at all, and gives a weird dll error.
I've tried recreating it as a couple different kinds of screens, but I followed this tutorial where they use a List and Details Screen in this way. The best I can come up with is that I'm experiencing a bug due to having VS 2012 and .NET 4.5 installed as well (don't ask me why). I'm just curious if this makes sense to anyone else, or if anyone else has been successful in creating these.
P.S. - I've also been unsuccessful in accessing the permissions I created by their name. I eventually figured out I could reference them by their index, but why is that necessary? I saw nothing about needing to add a reference to make this work.
partial void CustomerNew_CanRun(ref bool result)
{
result = User.HasPermission(Permissions.AllPermissions[1]);
}
OK, just to reassure you a bit (maybe), this is not normal, & I haven't seen this or heard of anything similar.
The fact that you can't access any of your permissions by name indicates something's definitely not right. You should be able to use Permissions.NameOfOneOfMyPermissions.
Starting from the basics, have you rebuilt the solution (note rebuilt, not just built)? What you'll often find with LightSwitch is that one error will cause a whole series of other errors to appear in the list, which can look quite daunting. When you fix the original error, the rest disappear. Double-click the error in the Error pane, & you should be taken to where the error is occurring. If you can't make sense of it, past the code here & I'll see if I can figure out what's wrong.
You'll also need to check that there isn't some code left over from any previous attempt at following the tutorial. Try adding a query with a different name, & see if the error has the new name in it, or the old one.

How to detect syntax errors when debugging Firefox extensions

When working on a large Firefox plugin, I sometimes accidentally make a syntax error not caught by my editor. When this happens, then calling
Components.utils.import("resource://modules/Foo.js")
will simply fail to import or return anything, without any kind of helpful hint as to where in the file to look for the syntax error. Is there any way I can get Firefox to give me some kind of clue when my imports fail?
EDIT: I fixed my own problem, which turned out to be that I was using code which had a global reference to the navigator object. What made this especially annoying was that the code would work when loaded in the browser (as Wladimir suggested below), but would still fail when importing in my extension.
Eventually I resorted to a sort of manual binary search: I'd delete half of the file and then see whether the import still failed. If so, then I'd delete half of what remained and repeat. As soon as it didn't fail, I had a more precise notion of where the problem was, which allowed me to either continue the binary search or scan the smaller area manually looking for the problem.
This is extremely time-consuming and I'd still appreciate any suggestions about how to speed up this debugging process.
The issue here is most likely the one described on https://developer.mozilla.org/en/Exception_logging_in_JavaScript and setting dom.report_all_js_exceptions preference to true should work. I must admit that I haven't tried that however because setting this preference makes the error console very noisy. Instead I use an ugly hack and load the module as a script in a local HTML file - this is enough to show me syntax errors and fortunately isn't something I need to do all too often (it is only an issue with syntax errors, runtime errors are reported as usually).

Package Manager Console can't find packages

I'm trying to
Install-Package NHibernate
but I'm getting "Package ... could not be found". If I open the PCM-options, the list of available sources is filled, but the green OK-symbol isn't there.
Can I use an alternative source different to the MS-Links-ID 206669 which is the default source in VS?
I had a similar problem last night. The two LinkID values I had were:
206669 and 19913
Meaning that the URL I pass into the Settings dialog were:
https://go.microsoft.com/fwlink/?LinkID=206669
https://go.microsoft.com/fwlink/?LinkID=19913
The 19913 did not yield anything in the list of packages available. I think this number came when I installed the package a few weeks (months?) ago and gave up the first time.
The 206669 seemed to work. I could see the expected packages, but I could only install some packages, not all. For example, I could get Castle, but not Log4net. I could get AutoMapper, but not RestSharp. The ones that failed simply said "operation timeout".
I'm glad that I'm not the only one to see these issues - I did not see anything about this in the FAQs or documentation.
It seems that the LinkID and that URL are a weakness in the framework. I would like to see more discussion about that these values mean, and why there seem to be now at least 3 different values to use. How about a more clear link like www.nuget.com?
And how about a message showing more clearly why an operation times out on one package but not another.
NuGet has great promise, and many people swear by it. I want to get on board, but hiccups like this can really stop people from adopting it.
Cheers,
Daniel

MSVC++ 6.0: Fatal error C1509 "Too many exception handler states in function"

I'm working on a project that uses a lot of generated files for a GUI. After some recent updates, I get this error. If I remove some controls (the new ones or some old ones, doesn't seem to matter which), the error goes away. I've temporarily worked around the problem, but I'd like a more permanent solution. I've seen mentioned on the web that there's a fix for this in VS.Net 2003, but I can't find anything for 6.0. Does anyone know of a hotfix or service pack that fixes this?
This is an example of why I became a maverick and developed another way of programming UIs.
It has none of these crazy issues, automatically does dynamic changes on-the-fly, and reduces source code by a large factor.

Resources