VBA Rubberducking

Duga

00:00

RELOAD!

[bruglesco/fleet-command] 2 commits. 82 additions. 92 deletions.

[rubberduck-vba/Rubberduck] 9 commits. 5 opened issues. 2 closed issues. 15 issue comments. 233 additions. 195 deletions.

[rubberduck-vba/Rubberduck] Stonovsky opened issue #4588: Wrong Interface implementation

> Sometime it happens, that there is wrong interface implementation. I tried to figure it out but I'm not sure when (it may be after renaming Interface class using Rubberduck). Please see following code with wrong implementation:

```
Option Compare Database
Option Explicit
Implements clsIWstawTowarDoFrm

Private Property Get clsIWalidacja_CzyWalidacjaPozytywna() As Boolean
Err.Raise 5 'TODO implement interface member
End Property

Private Property Get clsIWalidacja_PowodBrakuWali

[rubberduck-vba/Rubberduck] comintern assigned comintern to issue #4393: Missing Open Designer menu

[rubberduck-vba/Rubberduck] comintern added label bug to issue #4588: Wrong Interface implementation

[rubberduck-vba/Rubberduck] comintern added label feature-refactorings to issue #4588: Wrong Interface implementation

Duga

01:02

[rubberduck-vba/Rubberduck] comintern created comment on issue #4586: False positive rename conflict

> Replicated, and my initial instinct from [chat](https://chat.stackexchange.com/transcript/message/47841440#47841440) appears to be correct - `FindNewDeclarationNameConflicts` is picking up the `Item` member of `Collection` as being a conflict:

![screenshot from 2018-11-30 18-51-51](https://user-images.githubusercontent.com/8944005/49322072-80243980-f4d1-11e8-8cfd-ec262272fcd1.png)

Without diving too deep into the code, my suspicion is that the default member access on this line...

`Se

[rubberduck-vba/Rubberduck] comintern added label feature-refactorings to issue #4586: False positive rename conflict

[rubberduck-vba/Rubberduck] comintern added label refactoring-rename to issue #4586: False positive rename conflict

[rubberduck-vba/Rubberduck] comintern added label bug to issue #4586: False positive rename conflict

[rubberduck-vba/Rubberduck] comintern added label difficulty-02-ducky to issue #4586: False positive rename conflict

[rubberduck-vba/Rubberduck] comintern added label has-workaround to issue #4586: False positive rename conflict

Comintern

@Duga Just out of curiosity, could we simply try to bind the new name as whatever type it's declared as? If it's a potential naming collision, it would bind to something - if not, it wouldn't.

Vogel612

01:48

@this yea, it's hard, if not impossible to unify these...

Duga

[rubberduck-vba/Rubberduck] comintern opened issue #4589: RD should inspect for potential Worksheet_Change event re-entry.

> This might seem like an oddly specific inspection, but I'm finding myself dropping comments like this one on SO with increasing frequency:

> You're getting into an infinite loop because you're changing the worksheet from inside the Change event. Use Application.EnableEvents = false before you make changes, then set it back to True when you're done.

The code below will lock up Excel the first time it executes:

```
Private Sub Worksheet_Change(ByVal Target As Range)
Target.Value =

this

blech VBA makes arrays far too hard to work with.

and arrays of UDTs are extra painful.

Duga

[rubberduck-vba/Rubberduck] Vogel612 added label enhancement to issue #4589: RD should inspect for potential Worksheet_Change event re-entry.

[rubberduck-vba/Rubberduck] Vogel612 added label difficulty-04-quackhead to issue #4589: RD should inspect for potential Worksheet_Change event re-entry.

[rubberduck-vba/Rubberduck] Vogel612 added label code-path-analysis to issue #4589: RD should inspect for potential Worksheet_Change event re-entry.

[rubberduck-vba/Rubberduck] Vogel612 added label feature-inspections to issue #4589: RD should inspect for potential Worksheet_Change event re-entry.

[rubberduck-vba/Rubberduck] Vogel612 added label up-for-grabs to issue #4589: RD should inspect for potential Worksheet_Change event re-entry.

Vogel612

@Duga that one is cool, but I'm pretty sure that it has tons and tons and tons of edge-cases. Sooo ... I gave it a preemptive 04-quackhead

Comintern

Yeah, that's not a bad call. Heat of the moment.

this

01:59

I would be impressed if we can infer that from code analysis alone

I doubt there's anything that would indicate that this will cause an infinite loop. That's a result of implentation.

Comintern

I'd leave ambiguous cases alone, but there are plenty where it wouldn't be that hard to pick them out - i.e., the trivial example.

Interesting. I reflected the resources per #4534 and got 18 instead of the 16 we're currently setting. Wonder what's missing...

this

i'm wondering - excel simply just locks up? no runtime error?

Vogel612

well... it doesn't stackoverflow because it's not really recursion that's happening

so you just saturate the event queue

this

I asked because in Access if you do something like:

Private Sub Form_BeforeUpdate(Cancel As Integer)
  Me.Dirty = False ' save the record
End Sub

Comintern

Hmmm.... missing 3: Rubberduck.Resources.CommandBarIcons, Rubberduck.Resources.Settings.AutoCompletesPage, and Rubberduck.Resources.CodeExplorer.CodeExplorerUI. Any of those that shouldn't be set?

this

02:12

this throws a runtime error immediately on the attempt to save a record in process of saving a record.

obviously Access has some checks to ensure that this doesn't happen.

Comintern

Excel's Range don't behave like a recordset though. Excel != database.

this

hmm. aren't we supposed to use icons from CodeExplorerUI

but obivously it's working so... probably not?

Comintern

If it's just icons, those aren't localized.

this

@Comintern That's an Access form event.

Comintern

Right, but isn't that bound to a recordset?

this

02:15

not sure I can do the equivalent using DAO alone.

(no events with DAO model)

Vogel612

@Comintern more correctly, they don't have multiple cultures

Comintern

Right.

Vogel612

I don't think there's a drawback to including uncultured resource classes

this

uh stupid question - why would you localize an icon?

Vogel612

they fallback to neutral anyways

Comintern

02:16

So no objection to setting Culture = culture on them?

@this If it has a word in it.

this

yes figured that out too slowly

Comintern

Seems like a better argument for a more abstract icon though.

OK, how's this look?

        private static void LocalizeResources(CultureInfo culture)
        {
            var localizers = AppDomain.CurrentDomain.GetAssemblies()
                .SingleOrDefault(assembly => assembly.GetName().Name == "Rubberduck.Resources")
                ?.DefinedTypes.SelectMany(type => type.DeclaredProperties.Where(prop =>
                    prop.CanWrite && prop.Name.Equals("Culture") && prop.PropertyType == typeof(CultureInfo) &&
                    (prop.SetMethod?.IsStatic ?? false)));

this

just a question - why not use CW?

figured that since CW is already reflecting, might as well piggyback it, dunno.

Comintern

Does CW do static property injection one time only?

Vogel612

we don't have static propery injection set up

this

02:21

Also, this assumes that all resources file will ever be in the Rubberduck.Resources - do we need that to be always true?

Comintern

I would think so.

Vogel612

@this nah we don't but reflecting over the Attribute that the generator adds should add more typesafety

especially for the plugin system later

we want plugins to be able to localize

I'd also prefer CW. do we need LocalizeResources be static?

Comintern

I'd assume that would be on the IPluginInterface or whatever.

this

yeah, I guess that if we do wnat to assume it to be true, may need a simple analyzer to check that there are no resource files in other assemblies.

Comintern

Otherwise we couldn't "enforce" localization.

Vogel612

02:23

@this at least none that don't have the default culture

this

analyzer it is then.

Vogel612

@Comintern that makes sense, yeah

Comintern

I'm going to call that good for now then.

this

:+1:

Vogel612

Question: Can we not make use of Thread.CurrentUICulture?

Duga

02:25

[rubberduck-vba/Rubberduck] comintern assigned comintern to issue #4534: Resource/resx-generated classes are hard-coded in App.cs

Comintern

@Vogel612 The culture parameter is coming from the user settings.

It normally gets updated with a settings change event handler, but we need to set it to the correct setting at startup.

Vogel612

neither is something that rules out using Thread.CurrentUICulture

Comintern

I'm trying to remember if the settings default to that on first load or not. I was thinking they did.

Vogel612

oh well, if it works, I ain't gonna complain :)

high time to go to bed though.

Comintern

#4009 has me completely stumped. There is absolutely no reason for LoadLibrary to fail on a COM type that was obtained via interop.

Vogel612

02:29

Thanks for the hard work folks :)

Comintern

'night

this

hmm here's a interesting one.

sorry about the artifact of screenshot - but you can see the quickinfo saying it.Foo = Nothing

Comintern

Wut?

this

IKR?

Comintern

I wonder if it's just getting confused by the dereference operator in the tooltip.

this

02:35

@Comintern could it be related to the case #3837?

Comintern

it.Foo is uninitialized.

this

Correct. That's the most likely explanation

it just made a bad assumption that uninitialized => it's a object, so it's Nothing!

(whoops)

Comintern

Looking at the log for that one now.

this

it's a bit different because in that case, he couldn't even load the RD whereas this guy apparently just can't run tests

Comintern

That error looks like it resolved the types correctly.

this

02:37

That was when Carlos commented about how shims fixes this issue.

Comintern

The new one is the Interop layer faceplanting on a LoadLibrary call.

I wonder if the OP just hadn't recompiled the project on the machine or had some registry errors or something.

this

hmm. it's loading our own library, right?

Comintern

No, it was trying to get type information for the VBE in order to wrap it if I'm reading the log right.

> This operation failed because the QueryInterface call on the COM component for the interface with IID '{0002E166-0000-0000-C000-000000000046}' failed due to the following error: Error loading type library/DLL.

That's the GUID for the VBE.

this

yeah.

Comintern

That originated in Microsoft.Vbe.Interop.VBE

I'm not really sure how that happens - we're not even specifying the GUID - the Interop is.

this

02:41

but where's the LoadLibrary?

Comintern

It's internal - probably in the binding of the interop assembly.

this

ok so not by our code.

Comintern

^

this

Maybe take up his offer for skype debugging?

Comintern

I'm not sure I'd know what to look for.

We don't ship an interop do we?

this

02:44

I don't think so, but see, that's why I thought of 3837

because if (and that's a big if), he loads a different add-in and that changes the registration, that might cause the faceplant

Comintern

Wait a sec, does Office require that the VBE is registered? If it's early bound it shouldn't need to be.

this

which would be totally weird because 3837 (and as per Carlos' description) this kind of thing happens at startup not at runtime

TBH, if you achieved to launch Office without VBA registered.... #AchievementUnlocked

Comintern

Ahhhh... I think you might be onto something.

I'm gonna ask if he has any other addins loaded.

Duga

[rubberduck-vba/Rubberduck] comintern created comment on issue #4009: Excel 2016: unexpected error

> @tdalon - I was taking a look at this and was trying to come up with reasons why LoadLibrary would be failing. Do you remember if you had any other add-ins loaded when you got this error?

This may be related to #3837 (need COM shims).

[rubberduck-vba/Rubberduck] comintern added label bug to issue #4584: Excel crash on exit

[rubberduck-vba/Rubberduck] comintern added label difficulty-03-duck to issue #4584: Excel crash on exit

[rubberduck-vba/Rubberduck] comintern added label regression to issue #4584: Excel crash on exit

[rubberduck-vba/Rubberduck] comintern created comment on issue #4584: Excel crash on exit

> Can you dig the WER report out of the Event Log for one of these crashes and post the details here? The proximate cause is this:

```

2018-11-30 08:29:34.4297;TRACE-2.3.0.4235;Rubberduck.VBEditor.WindowsApi.SubclassManager;Releasing subclass for hWnd 0x00581520.;
2018-11-30 08:29:34.4609;ERROR-2.3.0.4235;Rubberduck._Extension;System.Exception: RemoveWindowSubclass Failed
at Rubberduck.VBEditor.WindowsApi.SubclassingWindow.ReleaseHandle() in C:\projects\rubberduck\Rubberduck.VBEEditor

Duga

03:39

[rubberduck-vba/Rubberduck] comintern created comment on issue #4408: Test Explorer>Run>Selected test no longer available

> I can't replicate this. Are you referring to Run->Selected Test from the toolbar?

[rubberduck-vba/Rubberduck] comintern assigned comintern to issue #4259: ImplicitByRefModifier Debug.Assert failed; a module is null

this

great....

> The menu option Analyze > Run Code Analysis applies only to static code analysis. Additionally, on a project's Code Analysis property page, the Enable Code Analysis on Build and Suppress results from generated code checkboxes apply only to static code analysis. They have no effect on Roslyn analyzers

Comintern

I should switch back to working on the references dialog before I forget WTH I was doing.

Duga

[rubberduck-vba/Rubberduck] comintern synchronized pull request #4581: Various quick bug fixes for 2.3 patch release

Duga

04:11

[rubberduck-vba/Rubberduck] build for commit 27f3dab8 on unknown branch: AppVeyor build succeeded

Duga

04:22

[rubberduck-vba/Rubberduck] bclothier synchronized pull request #4582: Deployment enhancements - (WIP - testing AV)

[rubberduck-vba/Rubberduck] build for commit 3f223258 on unknown branch: AppVeyor build failed

BUILD FAILURE!

Duga

04:44

[rubberduck-vba/Rubberduck] bclothier synchronized pull request #4582: Deployment enhancements - (WIP - testing AV)

Duga

05:03

[rubberduck-vba/Rubberduck] build for commit d8845790 on unknown branch: AppVeyor build failed

BUILD FAILURE!

David

Hi there, I'm a college student majoring in computer science with a entry level background in C/C++ development. I was wondering if there was any way I could learn from and contribute to your guys' project? I was referred here by a user by the name of IvenBach.

this

Welcome! We always need help. :-)

Duga

[rubberduck-vba/Rubberduck] bclothier synchronized pull request #4582: Deployment enhancements - (WIP - testing AV)

this

Have you ever done any VBA, David?

David

No unfortunately not, but I am very willing to learn!

this

05:16

Right I only asked because that's our primary focus - helping those who write in VBA to be more productive and be a better coder.

So it helps a bit to know about what they usually do in VBA but that's not really required for working in Rubberduck itself, which is mainly C#

Have you seen the contributing pages?

David

Gotcha, I would be starting from scratch with VBA personally. I did take a look at your guys' GitHub page although I've personally never used the site as my courses never made use of it.

this

(we have two -- here and here )

Are you familiar with git?

David

Not particularly sorry

this

No worries. that's what GitHub is based on and that's how we'll be contributing.

David

Gotcha sounds good

this

05:21

Visual STudio has some git functionality built in

but some here use command line, others like me use a plugin instead (TortoiseGit + TGit)

Regardless of what you do use, you will at least need a passing familiarity with basics of git - starting with git clone so that you can get the project set up on your computer

David

Oh ok. For the courses I've taken, I used Visual Studio so if I could use that, that would be great, but honestly whatever works best.

this

Yeah, you would need to use VS 2017

David

Yep, I have that installed.

this

Great! and we do have a wiki on the GH which provide some starting point - it may help to read through and get idea of what we do and how.

Once you've got your codebase set up, you will probably want to play with it - look for any open issues that has difficulty 01 or 02 and maybe is up for grabs and just hack away.

Do not worry about doing it wrong; that's why we do code reviews all the time - when you put out a PR we would read it through and give feedback or suggestions until it's a quality that's suitable before we merge it into the project.

and of course, you can use this chat room to ask any questions so you can get good guidance when you're hacking

Duga

[rubberduck-vba/Rubberduck] bclothier edited pull request #4582: Deployment enhancements - (WIP - testing AV)

[rubberduck-vba/Rubberduck] bclothier review_requested pull request #4582: Deployment enhancements - (WIP - testing AV)

[rubberduck-vba/Rubberduck] bclothier added label PR-Status: Review Requested to pull request #4582: Deployment enhancements - (WIP - testing AV)

[rubberduck-vba/Rubberduck] bclothier added label PR-Status: WIP to pull request #4582: Deployment enhancements - (WIP - testing AV)

[rubberduck-vba/Rubberduck] bclothier edited pull request #4582: Deployment enhancements

David

05:28

Gotcha, I was just making an account. I will get to installing the codebase now. Thanks for all your assistance!

this

Awesome, and have fun. I hope to see you around here. :-)

Duga

[rubberduck-vba/Rubberduck] build for commit 2b2a03c3 on unknown branch: AppVeyor build succeeded

David

Hey sorry how do I go about building the solution in VS?

this

05:50

Do you see a build menu?

Build -> Build Solution

David

No I don't see the option. I'm on the VS start page where you'd make a project/sln

IvenBach

@David Welcome to the pond.

David

Hi thanks!

Was just getting help trying to get started using VS

IvenBach

this along with the other ducks helped me get started with VS too.

David

06:07

Nice. I'm still just trying to figure out how to implement the build in my VS but I can't even so much as find a build option so perhaps I have a different type of VS downloaded?

I haven't written code on anything outside of creating an empty project for C/C++ or using the arduino compiler so I'm not quite sure how this works.

IvenBach

You'll need to get a copy of the repository, aka repo.

Since the code base is hosted on GitHub we usually use that to create a new branch and to clone a new repo.

If you don't have GH installed you can download a zip of the repo and extract it to wherever is convenient for you.

Once you have the repo cloned or extracted open the Rubberduck.sln file in VS. That will add Build to the menu.

David

oh ok gotcha. I downloaded GH because I couldn't find any options in VS

Where would I get a zip of the repo?

IvenBach

Look it the top right of this page. You should see rubberduckvba.com | github.com/rubberduck-vba/Rubberduck

github link will take you driectly to the github repo.

You'll see a green button that says "Clone or download".

David

oh ok so I downloaded it from the homepage

IvenBach

extract that zip to wherever you want to work on it.

When you do open the folder and it will contain the solution file, that's the one you'll want to navigate to and open with VS.

David

06:21

ok gotcha, just finished opening the sln

IvenBach

There are 19 projects within the solution. Each one AFAICT focuses on a specific part.

chat.stackexchange.com/rooms/info/14929/… Are various saved conversations that could be helpful in familiarizing with some parts.

David

ok and this had told me to check the issues list for different level contributions as well so should I start there?

IvenBach

How to Git, and Git branching. Both should help with Git knowledge.

Any and all issues are up for grabs. The 01-duckling and 02 issues are easier

The difficulties are best guesses.

David

Ah makes sense.

IvenBach

We’re all imposters here. Some fake it better than others.

@David github.com/rubberduck-vba/Rubberduck/labels/… is another tag you can filter on

David

06:36

oh perfect. That'll help narrow down what I can try to do

I'll take a more in-depth look when I get back from the gym. Thanks!

IvenBach

Oh and don't forget to star us on GitHub when you register.

@MathieuGuindon we got another incoming star from @David, our soon to be contributor.

Duga

[rubberduck-vba/Rubberduck] darxlmao starred us

3

David

gotcha :)

IvenBach

@Duga And there's our 700th star.

IvenBach

07:19

@David Sent you an email.

Night pond.

Duga

10:49

[rubberduck-vba/Rubberduck] InExSu opened issue #4590: All code in unit tests

> Thanks to your RD, i start learning to write a visual BASIC code suitable for testing.

For myself, I made a visual basic conversion of the code in the clipboard into templates for insertion into the RD module.

Now it looks like this: I wrote a macro consisting of several functions and I want to debug them.
I select all the written code with my **hands**, copy the code into the buffer with my **hands** and launch from my add-in, , again with my **hands**, my macro, which pulls out the su

Duga

11:19

[rubberduck-vba/Rubberduck] Vogel612 added label feature-templates to issue #4590: All code in unit tests

[rubberduck-vba/Rubberduck] Vogel612 added label enhancement to issue #4590: All code in unit tests

[rubberduck-vba/Rubberduck] Vogel612 added label up-for-grabs to issue #4590: All code in unit tests

[rubberduck-vba/Rubberduck] Vogel612 added label feature-unit-testing to issue #4590: All code in unit tests

[rubberduck-vba/Rubberduck] Vogel612 created comment on issue #4590: All code in unit tests

> I very much like this idea, something like a "generate test for member". We do not currently have any internal infrastructure supporting this, but it's definitely something that I consider in scope.

I've labelled this with [feature-templates], since we can probably reuse some infrastructure changes from the templates feature in something like this.

We'll need a way to clearly associate TestModules to Modules. This can get tricky comparatively fast. If in doubt, this should prompt for a m

Duga

12:00

[rubberduck-vba/Rubberduck] InExSu created comment on issue #4590: All code in unit tests

> Then I suffered :-)

I will try to implement in my code, which will generate a unit test based on the types of arguments received and returned by the function being tested,

The test will prepare the data for this function, start the function and check the return value or whatever I want.

Crazy! It will be mega cool! :-)

Duga

12:57

[rubberduck-vba/Rubberduck] Vogel612 commented on Rubberduck.Deployment.Build/RubberduckPostBuildTask.cs of pull request #4582: Deployment enhancements

[rubberduck-vba/Rubberduck] Vogel612 commented on Rubberduck.Deployment/Rubberduck.Deployment.csproj of pull request #4582: Deployment enhancements

[rubberduck-vba/Rubberduck] Vogel612 commented on RubberduckBaseMetaProject.csproj of pull request #4582: Deployment enhancements

[rubberduck-vba/Rubberduck] Vogel612 commented on Rubberduck.Deployment.Build/RubberduckPostBuildTask.cs of pull request #4582: Deployment enhancements

[rubberduck-vba/Rubberduck] Vogel612 commented on RubberduckBaseMetaProject.csproj of pull request #4582: Deployment enhancements

[rubberduck-vba/Rubberduck] Vogel612 commented on Rubberduck.Deployment.Build/RubberduckPostBuildTask.cs of pull request #4582: Deployment enhancements

[rubberduck-vba/Rubberduck] Vogel612 commented on Rubberduck.Deployment.Build/Rubberduck.Deployment.Build.csproj of pull request #4582: Deployment enhancements

[rubberduck-vba/Rubberduck] Vogel612 commented on Rubberduck.Deployment.Build/RubberduckPostBuildTask.cs of pull request #4582: Deployment enhancements

[rubberduck-vba/Rubberduck] Vogel612 commented on RubberduckBaseProject.csproj of pull request #4582: Deployment enhancements

[rubberduck-vba/Rubberduck] Vogel612 commented on Rubberduck.Deployment.Build/RubberduckPostBuildTask.cs of pull request #4582: Deployment enhancements

[rubberduck-vba/Rubberduck] Vogel612 edited issue #4590: Generate UnitTest stubs for members and modules

Duga

14:44

[rubberduck-vba/Rubberduck] wqweto opened issue #4591: Parse error on mutiple WithEvents vars

> Grammar chokes on this
```
Private WithEvents a As VB.Form, WithEvents b As VB.Form
```
Btw, it just says `Parse Error` on the toolbar. Where can I see the actual error?

[rubberduck-vba/Rubberduck] Vogel612 added label difficulty-03-duck to issue #4591: Parse error on mutiple WithEvents vars

[rubberduck-vba/Rubberduck] Vogel612 added label bug to issue #4591: Parse error on mutiple WithEvents vars

[rubberduck-vba/Rubberduck] Vogel612 added label antlr to issue #4591: Parse error on mutiple WithEvents vars

[rubberduck-vba/Rubberduck] bclothier commented on RubberduckBaseProject.csproj of pull request #4582: Deployment enhancements

[rubberduck-vba/Rubberduck] Vogel612 created comment on issue #4591: Parse error on mutiple WithEvents vars

> The actual error message will be in the logfiles. Assuming this code compiles, we'd need to look at how `WithEvents` is treated in the grammar.

https://github.com/rubberduck-vba/Rubberduck/blob/87f612bc3b15fd0d7b0a321b6afc19749b001a29/Rubberduck.Parsing/Grammar/VBAParser.g4#L552-L554

The relevant grammar section is visible above. Quick, untested and unverified patch suggestion:

```diff
- variableStmt : (DIM | STATIC | visibility) whiteSpace (WITHEVENTS whiteSpace)? variableListStmt;

Duga

15:10

[rubberduck-vba/Rubberduck] MDoerner created comment on issue #4591: Parse error on mutiple WithEvents vars

> Regarding, seeing the parser arror(s), there should be a little red error icon right next to the refresh button stating Parser Error. When you click it, a window with the parser error(s) will open.

[rubberduck-vba/Rubberduck] MDoerner edited comment on issue #4591: Parse error on mutiple WithEvents vars

> Regarding, seeing the parser error(s), there should be a little red error icon right next to the refresh button stating Parser Error. When you click it, a window with the parser error(s) will open.

Duga

15:20

[rubberduck-vba/Rubberduck] Vogel612 commented on RubberduckBaseProject.csproj of pull request #4582: Deployment enhancements

Duga

16:22

[rubberduck-vba/Rubberduck] comintern created comment on issue #4591: Parse error on mutiple WithEvents vars

> The spec covers these in [5.2.3.1 Module Variable Declaration Lists](https://msdn.microsoft.com/en-us/library/ee177381.aspx)

> module-variable-declaration-list = (withevents-variable-dcl / variable-dcl)
> *( "," (withevents-variable-dcl / variable-dcl) )

Our grammer doesn't make a distinction between variable declarations in modules and procedure bodies (and we're already parsing non-compilable declarations inside procedures) but this syntax is clearly defined as legal.

Bas

Mathieu Guindon

16:33

@Duga ha, it only took 4 years to hit that one!

Comintern

IKR? Thinking difficulty-03-duck is an overstatement. The fix looks fairly trivial.

he says before running unit tests...

Mathieu Guindon

@Comintern agreed

Comintern

There was exactly one line of code that needed to be changed in the resolver.

Duga

[rubberduck-vba/Rubberduck] retailcoder added label difficulty-02-ducky to issue #4591: Parse error on mutiple WithEvents vars

[rubberduck-vba/Rubberduck] retailcoder removed label difficulty-03-duck from issue #4591: Parse error on mutiple WithEvents vars

Mathieu Guindon

@Comintern that's a very good sign!

Vogel612

16:41

ya, first reaction to a grammar bug is slap a high-level difficulty on it, because the grammar is just so advanced.

didn't think it'd be that easy a fix

Mathieu Guindon

@Duga love the enthusiasm there!

Vogel612

unrelated: apparently I can't automate cleaning up old prereleases with the v4 of Github's API

this

@Vogel612 why would a project that's importing a base project that explicitly specifies C# 7.2 think it's still using 7.1?

Vogel612

conflicting definition somewhere else?

VS being obtuse?

Mathieu Guindon

@Vogel612 eh, mine is to deem it trivial, then get bitten by one or two subtleties that escaped me - then break 1600 tests, sweat bullets, freak out, then realize it was just a missing comma or something

4

this

16:45

i looked for that first. I didn't see any conflict.

looks like more the latter.

Comintern

...and all green. Just needs a test covering the change and I think it's GTG.

Duga

[rubberduck-vba/Rubberduck] comintern assigned comintern to issue #4591: Parse error on mutiple WithEvents vars

Comintern

It's going to take me longer to figure out where the stupid test belongs than to fix the thing.

Mathieu Guindon

lol

GrammarTests?

this

i think the test project is overdue for a split

Comintern

16:56

I'm thinking a test there for the grammar, then maybe another one to make sure that the variable declarations are correct for the resolver.

this

it's getting a bit... impractical to build that test project just to test it.

Vogel612

@this it shouldn't really have been a single monolithical test project in the first place, IMO

that being said: it's probably going to make running tests on AV somewhat more complicated

this

#LazyFTW

I guess AV doesn't do globbing?

Vogel612

in some places it does, in some it doesn't...

We're running the tests in an explicit test_script anyways, so AV doesn't have much in the way of globbing to help us, I think

so the more interesting question is: does Powershell support globbing?

this

Gee. Even this doesn't work: <MSBuild Projects="$(SolutionDir)\RubberduckMeta.sln" />

Omit the path and it wigs about not finding the solution file.

Powershell in theory should be able to yeah.

Vogel612

16:59

hmm ... did you try specifying any Target?

this

  <Target Name="EnsureMetaProjectsBuild" BeforeTargets="PrepareForBuild"
          Condition="!Exists('$(SolutionDir)RubberduckCodeAnalysis\bin\RubberduckCodeAnalysis.dll')">
    <Message Text="Missing the required meta projects. Building the meta projects..." Importance="high" />
    <MSBuild Projects="$(SolutionDir)\RubberduckMeta.sln" />
    <Message Text="Built the meta projects" Importance="high" />
  </Target>

Is that what you mean?

Vogel612

<MSBuild Projects="$(SolutionDir)\RubberduckMeta.sln" Targets="CoreCompile" />

this

oh

hmm. ok that might be something - it failed, though.

17>  Missing the required meta projects. Building the meta projects...
17>  C:\GitHub\Rubberduck\Rubberduck\RubberduckMeta.sln.metaproj : error MSB4057: The target "CoreCompile" does not exist in the project.
17>
17>  Done building project "RubberduckMeta.sln" -- FAILED.

Vogel612

this msdn link is somewhat enlightening...

Let's try Targets="RubberduckCodeAnalysis:CoreCompile" just for testing

this

didn't work, either; said the target not found

> Targets Optional String parameter.

Specifies the target or targets to build in the project files. Use a semicolon to separate a list of target names. If no targets are specified in the MSBuild task, the default targets specified in the project files are built. Note: The targets must occur in all the project files. If they do not, a build error occurs.

this

17:26

welp, that's interesting.

  <Target Name="EnsureMetaProjectsBuild" BeforeTargets="PrepareForBuild"
          Condition="!Exists('$(SolutionDir)RubberduckCodeAnalysis\bin\RubberduckCodeAnalysis.dll')">
    <Message Text="Missing the required meta projects. Building the meta projects..." Importance="high" />
    <MSBuild Projects="$(SolutionDir)\RubberduckMeta.sln"
             Targets="Build"
             Properties="Configuration=Release">
      <Output TaskParameter="TargetOutputs"
              ItemName="MetaAssemblies" />

outputs:

1>Target EnsureMetaProjectsBuild:
1>  Missing the required meta projects. Building the meta projects...
1>  Target ValidateSolutionConfiguration:
1>    Building solution configuration "Release|AnyCPU".
1>  Built the meta projects

except.... it didn't actually build anything.

this

17:39

Ok, for now I'll put aside the solution; I put it in a comment in case anyone else want to try and fix it. I came up with a bit more dry approach for the msbuild so we only specify it once using a properties that's an array of the individual metaprojects.

Duga

17:53

[rubberduck-vba/Rubberduck] bclothier commented on Rubberduck.Deployment.Build/RubberduckPostBuildTask.cs of pull request #4582: Deployment enhancements

Duga

18:16

[rubberduck-vba/Rubberduck] bclothier edited pull request #4582: Deployment enhancements

[rubberduck-vba/Rubberduck] comintern edited comment on issue #4591: Parse error on mutiple WithEvents vars

> The spec covers these in [5.2.3.1 Module Variable Declaration Lists](https://msdn.microsoft.com/en-us/library/ee177381.aspx)

> module-variable-declaration-list = (withevents-variable-dcl / variable-dcl)
> *( "," (withevents-variable-dcl / variable-dcl) )

Our grammar doesn't make a distinction between variable declarations in modules and procedure bodies (and we're already parsing non-compilable declarations inside procedures) but this syntax is clearly defined as legal.

Bas

Comintern

@Duga No, I cannot spell "grammar" correctly. Ever.

Duga

[rubberduck-vba/Rubberduck] comintern synchronized pull request #4581: Various quick bug fixes for 2.3 patch release

[rubberduck-vba/Rubberduck] bclothier commented on Rubberduck.Deployment.Build/RubberduckPostBuildTask.cs of pull request #4582: Deployment enhancements

[rubberduck-vba/Rubberduck] bclothier commented on RubberduckBaseMetaProject.csproj of pull request #4582: Deployment enhancements

[rubberduck-vba/Rubberduck] comintern created comment on issue #4587: IgnoreOnce QuickFix puts the annotation under the target

> I can't replicate this.

I'm guessing that it if it's repeatable in the source module it has more to do with the layout of the entire module than anything else. Between this issue and #4586 I already have a workbook with a good chunk of your project stubbed out now, so it doesn't appear to be due to how it's parsing.

Is it possible that the quick-fix was run after another quick-fix that for some reason didn't refresh the parse?

[rubberduck-vba/Rubberduck] bclothier synchronized pull request #4582: Deployment enhancements

[rubberduck-vba/Rubberduck] bclothier edited pull request #4582: Deployment enhancements

[rubberduck-vba/Rubberduck] bclothier commented on Rubberduck.Deployment.Build/RubberduckPostBuildTask.cs of pull request #4582: Deployment enhancements

[rubberduck-vba/Rubberduck] build for commit b8b0b7d2 on unknown branch: AppVeyor build succeeded

[rubberduck-vba/Rubberduck] bclothier commented on Rubberduck.Deployment.Build/RubberduckPostBuildTask.cs of pull request #4582: Deployment enhancements

[rubberduck-vba/Rubberduck] comintern created comment on issue #4587: IgnoreOnce QuickFix puts the annotation under the target

> If anyone wants to play with this, I have it set up in a test workbook that I dropboxed here:

https://www.dropbox.com/s/uw9szelj6vwyrux/4586-7.xlsm?dl=0

[rubberduck-vba/Rubberduck] bclothier commented on Rubberduck.Deployment.Build/RubberduckPostBuildTask.cs of pull request #4582: Deployment enhancements

[rubberduck-vba/Rubberduck] bclothier commented on RubberduckBaseProject.csproj of pull request #4582: Deployment enhancements

[rubberduck-vba/Rubberduck] comintern created comment on issue #4586: False positive rename conflict

> There's a link to a workbook with the repro case on #4587

this

18:44

  No C++ build tools found; using tlbexp.exe to generate TLBs.
  Processing Rubberduck.dll...
  Compiling with midl...

I feel so smort.

Duga

[rubberduck-vba/Rubberduck] bclothier synchronized pull request #4582: Deployment enhancements

Comintern

Hmmm... #4588 is going to be tricky as hell.

I think the problem is that when you put Implements InterfaceToBeImplemented at the top of a class, it isn't able to compile until after the interface implementation is there.

this

Yeah, the compile before parse would definitely get in the way

Comintern

That means if you cancel the parse due to uncompilable code, it doesn't have the parse tree refreshed.

I got a weird partial repro on it though. I deleted the implementation, dismissed the compile warning, and when I did a rename refactor it re-implemented the interface I'd deleted.

this

fun.

Comintern

18:55

...and now I can't repro my weird repro. Nice.

Actually, I think the solution might be to just check to see if the module or the interface is dirty before enabling the refactoring.

Duga

[rubberduck-vba/Rubberduck] build for commit d8cdb8bc on unknown branch: AppVeyor build succeeded

this

IIRC, we have the compile before parse in order to avoid parser errors

Comintern

I'm going to see if I can sidestep it completely.

this

or maybe it was to avoid the false positives

since our grammar is more liberal than the actual compiler

Comintern

Wait, wut? We're already testing that:

    return targetInterface != null && targetClass != null
        && !_state.IsNewOrModified(targetInterface.QualifiedModuleName)
        && !_state.IsNewOrModified(targetClass.QualifiedModuleName);

this

19:00

e.g. an Implements without members wouldn't make parser bat an eyebrows but compiler flips out.

Comintern

^ And the refactoring would be pointless if it wasn't.

I'm almost thinking that the UI for that refactoring is wrong.

this

seems it'd be easier if parser can tolerate errors and refactoring will refuse to refactor blocks that contains an error node.

right now parser just wigs out at first error.

Comintern

What should happen is that if you're in a class module and select the refactoring, RD would prompt you with a list of available interfaces to implement.

Then it would add the Implements Foo at the top of the class along with the stubs.

this

yeah we should do that.

no uncompilable code, no problem.

BBL

Duga

19:17

[rubberduck-vba/Rubberduck] build for commit 13721382 on unknown branch: AppVeyor build succeeded

[rubberduck-vba/Rubberduck] comintern created comment on issue #4588: Wrong Interface implementation

> I was able to (kind of) get a repro on this. If you implement an interface, then delete the implementation, renaming the interface will re-implement it (with the new names) in the implementing class. This is apparently due to the rename refactoring not checking to see if a module is dirty before rewriting it with the new naming.

So @Stonovsky, just to be clear - did the implement interface command result in the wrongly named implementation, or did the rename add them?

I think part of the

Duga

19:59

[rubberduck-vba/Rubberduck] comintern opened issue #4592: Rename refactoring allows renamings that will rewrite dirty modules

> Given a project with a class module `Class1` and the following code in a standard module...
```
Public Sub Foo()
Dim bar As Class1
Set bar = New Class1
End Sub
```
...running the rename refactoring results in this if you rename `Class1` to `FooBar`:
```
Public Sub Foo()
Dim bar As FooBar
Set bar = New FooBar
End Sub
```
If you edit the sub and change the types to, say `Collection`...
```
Public Sub Foo()
Dim bar As Collection
Set bar = New Collection

[rubberduck-vba/Rubberduck] comintern added label bug to issue #4592: Rename refactoring allows renamings that will rewrite dirty modules

[rubberduck-vba/Rubberduck] comintern added label feature-refactorings to issue #4592: Rename refactoring allows renamings that will rewrite dirty modules

[rubberduck-vba/Rubberduck] comintern added label refactoring-rename to issue #4592: Rename refactoring allows renamings that will rewrite dirty modules

[rubberduck-vba/Rubberduck] comintern created comment on issue #4592: Rename refactoring allows renamings that will rewrite dirty modules

> Almost forgot - this might? be the root cause of #4588.

Duga

20:20

[rubberduck-vba/Rubberduck] comintern created comment on issue #4562: Inspecting non-compilable code wrongly sets busy state on inspection results toolwindow

> OK, gave this some more thought, and I think I'm fine with the inspection results toolwindow not giving feedback (spinning the ducks) in response to the parser state at all. The state of the parse should be obvious from the RD toolbar as well as the code explorer window (if it's open), so it would only look like a NOP if they're only looking at the inspection window. It wouldn't really be unexpected to have the ducks spin only while the inspection results are updating - i.e., the window is b

Duga

22:19

[rubberduck-vba/Rubberduck] mansellan created comment on issue #4555: High DPI causes severe IDE (and app running from IDE) graphics glitches

> Workaround:

1. Open Regedit, go to HKLM\Software\Microsoft\Windows\CurrentVersion\SideBySide. Add a DWORD of PreferExternalManifest with value 1.
2. Go to location of VB6.exe, add the following into a new file called VB6.exe.manifest:

```xml <?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<assembly xmlns="urn:schemas-microsoft-com:asm.v1" manifestVersion="1.0" xmlns:asmv3="urn:schemas-microsoft-com:asm.v3">
<asmv3:application>
<asmv3:windowsSettings xmlns="http://schemas

Vogel612

@Duga If that's confirmed to work, we should add the has-workaround label.

mansellan

@Duga if anyone else has VB6 on Windows Vista\7\8\8.1, would be good to try it there

worked for me on win10

if manifests don't work on earlier versions, there's no point us coding them as a fix. may as well stick with the 8.1 API

Duga

22:31

[rubberduck-vba/Rubberduck] mansellan edited comment on issue #4555: High DPI causes severe IDE (and app running from IDE) graphics glitches

> Workaround:

1. Open Regedit, go to HKLM\Software\Microsoft\Windows\CurrentVersion\SideBySide. Add a DWORD of PreferExternalManifest with value 1.
2. Go to location of VB6.exe, add the following into a new file called VB6.exe.manifest:

```xml <?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<assembly xmlns="urn:schemas-microsoft-com:asm.v1" manifestVersion="1.0" xmlns:asmv3="urn:schemas-microsoft-com:asm.v3">
<asmv3:application>
<asmv3:windowsSettings xmlns="http://schemas

[rubberduck-vba/Rubberduck] mansellan added label has-workaround to issue #4555: High DPI causes severe IDE (and app running from IDE) graphics glitches

Duga

22:52

[rubberduck-vba/Rubberduck] mansellan edited comment on issue #4555: High DPI causes severe IDE (and app running from IDE) graphics glitches

> There are a couple of workarounds available, depending on OS:

**For Windows 7 (and presumably Vista and 8, untested)**

1. Ensure DPI scaling mode is set to use Windows XP-style DPI scaling

**For Windows 10 (and presumably 8.1, untested)**

1. Open Regedit, go to HKLM\Software\Microsoft\Windows\CurrentVersion\SideBySide. Add a DWORD of PreferExternalManifest with value 1.
2. Go to location of VB6.exe, add the following into a new file called VB6.exe.manifest:

```xml <?xml ver

[rubberduck-vba/Rubberduck] mansellan removed label vb6-specific from issue #4555: High DPI causes severe IDE (and app running from IDE) graphics glitches

[rubberduck-vba/Rubberduck] mansellan created comment on issue #4555: High DPI causes severe IDE (and app running from IDE) graphics glitches

> Confirmed issue affects older versions of Office - removing vb6-specific tag

mansellan

@Duga Looks like manifests are not effective in Windows < 8.1, but neither are they needed. XP-style scaling seems to be all that's needed there (and googling suggests its the best option generally)

Transcript for