VBA Rubberducking

Duga

12:00 AM

RELOAD!

[rubberduck-vba/Rubberduck] 2 opened issues. 10 issue comments.

M.Doerner

12:32 AM

@Comintern Core contains the settings, too, which is a PITA.

Comintern

I think there was a historical reason for that, but I'm guessing it's probably not the case any more.

this

I think it had to do with loading the settings

but that should be in Rubberduck.Main?

I wonder if we can get away with moving them over to Rubberduck.SettingsProvider.

Comintern

I'm pretty sure it had to do with the dependency graph. That should be sorted out though.

SettingsProvider creates circular dependencies, no?

M.Doerner

In the current state, nothing in Parsing can depend on settings.

this

IDK. but even if that's the case, there's the Rubberduck.Resources which is explicitly meant to be on the bottom of the chain

(or top, however you view it)

though that's our 3rd junk drawer.... :(

Comintern

12:38 AM

The dependency problem is that some of the settings depend on strong types.

M.Doerner

That is purely for resources.

But, do they depend on types in Core?

this

I'm wondering if settings really ought to.

Comintern

Maybe. I'm giving it a quick check.

M.Doerner

Anyway, why should settings depend on anything?

this

yeah, it should be basically a model. One possible problem is the enumerations, though.

M.Doerner

12:42 AM

Ok, they could depend on some enums.

Comintern

Yep. That's all I found.

this

why is there a declaration.QualifiedModuleName and declaration.QualifiedName.QualifiedModuleName?

Comintern

Is that in main?

this

well, enums can legitimately be argued as being just data but...

no, core - OpenDesignerCommand

(main isn't supposed to touch anything from parsing)

Comintern

They're in Rubberduck.VBEditor.

this

12:48 AM

They being the enums?

Comintern

QMN etc.

this

ok, crossed the streams again -

but my question was is whether there's a difference between declaration.QualifiedModuleName and declaration.QualifiedName.QualifiedModuleName

Comintern

Ah, no. Looks like a shim.

public QualifiedMemberName QualifiedName { get; }
public QualifiedModuleName QualifiedModuleName => QualifiedName.QualifiedModuleName;

this

should have thought to look there

was a bit taken aback to see alternative routes, esp only one hop away

Comintern

I'm guessing it added as a convenience at some point.

this

12:57 AM

yeah i kind of can see it. declaration.QualifiedName.QualifiedModuleName would get quite winded very fast.

are we shooting to release 2.4 on sunday?

Comintern

IDK. Depends on how we feel about the lack of recent issues on the CE. TBH I was expecting more.

Do we have DL counts for 4431-pre and 4441-pre somewhere?

this

That'd be a question for @MathieuGuindon

Comintern

The RD GitHub page gets a crapload of daily visitors.

Mathieu Guindon

> Rubberduck.Setup.2.3.1.4441-pre.exe (4.02 MiB) - downloaded 60 times. Last updated on 2019-01-17

> Rubberduck.Setup.2.3.1.4431-pre.exe (4.02 MiB) - downloaded 49 times. Last updated on 2019-01-16

> Rubberduck.Setup.2.3.1.4414-pre.exe (4.01 MiB) - downloaded 198 times. Last updated on 2019-01-09

Comintern

Thanks!

Mathieu Guindon

1:09 AM

> Rubberduck.Setup.2.3.1.4308.exe (3.89 MiB) - downloaded 1,097 times. Last updated on 2018-12-10

this

oh nice!

1000 in a month?

well a bit over a month

but still.

Mathieu Guindon

Yeah, that's a first!

this

any idea last time we broke 1000?

Comintern

That's more installs than we get at work.

Granted, our software is significantly more expensive and is pretty specialized, but...

Mathieu Guindon

@this 1.4.3 was close to 5.5K IIRC, when 2.0 was launched

this

1:11 AM

how big was the gap between those 2?

Mathieu Guindon

a year=)

this

kind of a significant gap. :p

Mathieu Guindon

basically 1K/month never happened before AFAIK

this

I think back then in 1.4.3/2.0 (and I wasn't here then) you only had low hundreds of stars, didn't you?

now we have ~700, so more people are paying attention (?)

Comintern

It's interesting that our page view frequency graph matches the one we have at work. I wonder why Wednesday is "software install day".

Mathieu Guindon

1:13 AM

we've had 60-70 watchers for a while though

this

i think the majority don't really want the dev chatters.

watching would bring a lot of chatters that would go over the head

@Comintern because it's the hump day?

Mathieu Guindon

depends, GH introduced "releases only" watch mode this year, so in theory watching a repo isn't as annoying as it was

this

oh that's right.

might be still too new, though.

I'd wager the same crowd who don't want dev chatter either don't read GH news. :p

Mathieu Guindon

^

this

hmm. Why is the declaration.DeclarationType returning ClassModule while its a document (its ComponentType returns a Document.

Comintern

1:19 AM

Because a Document is a ClassModule.

this

yes, but the DeclarationType.Document is already OR'd

1L << 5 | ClassModule

Comintern

Well that explains the component type check I found in the CE...

this

the OpenDesignerCommand is just doing a simple equality. Felt wrong

I think the DeclarationType needs to be using HasFlag()

Comintern

Is DeclarationType.Document resolving correctly?

I'm trying to remember when we check that.

Oh, well there's yer problem:

public ClassModuleDeclaration(
          QualifiedMemberName qualifiedName,
          Declaration projectDeclaration,
          string name,
          bool isUserDefined,
          IEnumerable<IAnnotation> annotations,
          Attributes attributes,
          bool isWithEvents = false,
          bool hasDefaultInstanceVariable = false,
          bool isControl = false)
    : base(
          qualifiedName,
          projectDeclaration,
          name,
          DeclarationType.ClassModule,
          isUserDefined,

It's hard coded in the ctor.

Mathieu Guindon

Doesn't it make sense?

Or is DocumentModuleDeclaration inheriting ClassModuleDeclaration?

this

1:25 AM

well, the declaration i have is for a document.

this is cool, though:

switch (declaration.DeclarationType)
                {
                    case DeclarationType.ClassModule when qualifiedModuleName.ComponentType != ComponentType.Document:
                        return _projectsProvider.Component(qualifiedModuleName).HasDesigner;
                    case DeclarationType.ClassModule:
                        using (var app = _vbe.HostApplication())
                        {
                            return app != null && app.CanOpenDocumentDesigner(qualifiedModuleName);

didn't know about the when. Thanks, R#!

Mathieu Guindon

pattern matching ftw

Comintern

There isn't a DocumentModuleDeclaration. I'm presuming that they get created as ClassModuleDeclaration because they support interfaces.

(for some values of "support")

Mathieu Guindon

@Comintern makes sense

@Comintern ha yeah

this

yeah, that's "wrong", and probably explains the CE check you mentioned

I'm not going to fix that either. #PartOfTheProblem

Comintern

LOL. Are we even checking to see if it's a document in the resolver?

this

1:28 AM

hmm the QMN does have it right, though.

Comintern

I thought we were. And if so, it's an easy fix.

this

so maybe?

Mathieu Guindon

@this one day a VBA dev will say "oh nice, didn't know that - thanks Rubberduck!"

3

this

now that would be AWESOME

Comintern

@this it's a trivial fix.

this

1:30 AM

Rubberduck being awesome is a trivial fix?

Comintern

No, the declaration type for documents.

Add an isDocument parameter to the ctor.

this

do we really want to do that? :p

Comintern

Pass it from DeclarationResolveRunnerBase - it's around line 186.

this

feels like making the problem worse.

Comintern

Then fix the hundreds of broken tests.

:-D

this

1:31 AM

lol

Comintern

TBH, you should open an issue for it though. That's some unneeded tech debt.

It directly results in hitting the COM objects in some of the callers.

this

OK - i'll have a look at that once I have it working

this

1:47 AM

trying out the "trivial" fix

Hey, @Comintern want to guess how many failed tests?

1500?

0

Nice!

should we be scared instead?

Comintern

Hey, @this want to guess how many upcoming issues?

this

1:56 AM

666

Comintern

Let me do a quick sanity test on uses of that declaration type.

I do have the feeling this will break stuff. The more I think of it, the less coverage I think we have on the distinction between classes and documents.

this

That's why I'm scared - if no unit tests fail, it usually implies that there's a dearth of coverage

TBH, I haven't written any unit tests for the changes I made to OpenDesignerCommand. :(

Comintern

Documents were the bastard stepchild of classes for a long time.

this

yeah I can understant that.

we're only starting to break apart the egg.

Comintern

Probably safer to leave it out and open the issue. It'll need a pretty comprehensive static review.

this

2:02 AM

np - i'll stash that change.

IvenBach

@MathieuGuindon pfft. RD already did that for me with pretty much everything.

this

attaches teh debugger while unit testing is running

that looks like a good idea.

Duga

[rubberduck-vba/Rubberduck] bclothier assigned bclothier to issue #4708: "Open Designer" does not work for document modules

[rubberduck-vba/Rubberduck] bclothier created comment on issue #4722: Crash upon VBE resize

> @Greedquest is it possible to get a copy of your VBE registry, in particular the Dock key? I have a hunch we need that in order to reproduce.

[rubberduck-vba/Rubberduck] bclothier edited comment on issue #4722: Crash upon VBE resize

> @Greedquest is it possible to get a copy of your VBE registry, in particular the `Dock` key? I have a hunch we need that in order to reproduce.

I cannot reproduce, even though I made VBE window small then restart Excel.

Duga

2:41 AM

[rubberduck-vba/Rubberduck] bclothier assigned bclothier to issue #4719: Code Inspection: Incorrect recommendation to switch from ByRef to ByVal for array

[rubberduck-vba/Rubberduck] bclothier opened issue #4724: DeclarationType is not correct for documents

> That was discovering while fixing #4708 -- for documents, they are given `DeclarationType.ClassModule`. That is because of the document falling into the `default` case:

https://github.com/rubberduck-vba/Rubberduck/blob/dda930087260194b2e3935fac1171143b6f5cdf4/Rubberduck.Parsing/VBA/DeclarationResolving/DeclarationResolveRunnerBase.cs#L186-L213

The fix is trivial; we can just add a `isDocument = false` to the `ClassModuleDeclaration` and set the declaration type accordingly. The fix does

[rubberduck-vba/Rubberduck] bclothier added label bug to issue #4724: DeclarationType is not correct for documents

[rubberduck-vba/Rubberduck] bclothier added label technical-debt to issue #4724: DeclarationType is not correct for documents

[rubberduck-vba/Rubberduck] bclothier milestoned issue #4724: DeclarationType is not correct for documents

[rubberduck-vba/Rubberduck] bclothier added label difficulty-02-ducky to issue #4724: DeclarationType is not correct for documents

Duga

3:24 AM

[rubberduck-vba/Rubberduck] CHR-IS-B opened issue #4725: Indenter does not indent preprocessor commands within procedures

> **Rubberduck version information**
Version 2.3.0.4261
OS: Microsoft Windows NT 6.1.7601 Service Pack 1, x64
Host Product: Microsoft Office 2016 x86
Host Version: 16.0.4771.1000
Host Executable: EXCEL.EXE

**Description**
Preprocessor commands such as `#If` used within procedures are not indented.

**To Reproduce**
1. Place this in a code module:
```
Sub ReproduceBug()
#If 1 = 1 Then
MsgBox "Math still works!"
#Else
MsgBox "Match is broken."
#End If
End Sub
```
2. Run t

[rubberduck-vba/Rubberduck] CHR-IS-B added label bug to issue #4725: Indenter does not indent preprocessor commands within procedures

[rubberduck-vba/Rubberduck] comintern created comment on issue #4725: Indenter does not indent preprocessor commands within procedures

> Can you check the following 2 indenter settings that could effect this?

Force compiler directives to column 1 would cause this behavior if checked.
Indent compiler directives would (for the sample code above) cause this behavior if unchecked.

If the problem still persists, there is an Export link in the top right corner of the indenter settings form - click the link, and then append the resulting file to the issue.

Thanks!

[rubberduck-vba/Rubberduck] comintern added label feature-smart-indenter to issue #4725: Indenter does not indent preprocessor commands within procedures

Duga

3:46 AM

[rubberduck-vba/Rubberduck] bclothier created comment on issue #4645: False positive 'Parameter can be passed by value'

> I am unable to reproduce this. In an empty project, regardless whether it's a module, a class or a document, I cannot get the inspection to fire for the given code.

Is there more to the story than the original report is letting on?

Duga

5:53 AM

[rubberduck-vba/Rubberduck] bclothier assigned bclothier to issue #4677: Encapuslate Field is broken

Duga

6:21 AM

[rubberduck-vba/Rubberduck] bclothier assigned comintern to issue #4660: Clean up WPF leaks

[rubberduck-vba/Rubberduck] bclothier assigned bclothier to issue #4625: Return value not used false positive given object variable

Comintern

@Duga I suppose I should break out the profiler. ;-P

this

sorry - i assumed since you were on other WPF issues, and it might be similar enough to be rolled up

Comintern

No, I just missed that one.

Duga

[rubberduck-vba/Rubberduck] comintern synchronized pull request #4717: Inspection Result window performance overhaul (Round 1)

[rubberduck-vba/Rubberduck] comintern created comment on issue #3321: Spinning Ducks stop spinning between end of Parse, and refresh of Inspection Results

> This should be fixed when #4717 merges.

[rubberduck-vba/Rubberduck] comintern created comment on issue #2780: Code Explorer and Code Inspections windows have spinning/busy icon with 0 open projects

> This should be fixed when #4717 merges.

Comintern

@this we good to merge #4721? Hotfix is cooling down.

Duga

6:35 AM

[rubberduck-vba/Rubberduck] bclothier closed issue #4720: Auto folder annotation issue

[rubberduck-vba/Rubberduck] bclothier merged pull request #4721: Code Explorer folder hotfix 2 from comintern/Rubberduck/next into rubberduck-vba/Rubberduck/next

[rubberduck-vba/Rubberduck] comintern pushed commit 446f1e1c to next: Handle folder component sync correctly. Closes #4720

[rubberduck-vba/Rubberduck] comintern pushed commit 20731550 to next: Add annotation order tests.

[rubberduck-vba/Rubberduck] web-flow pushed commit 81a26265 to next

Merge pull request #4721 from comintern/next

Code Explorer folder hotfix 2

[rubberduck-vba/Rubberduck] bclothier closed issue #2780: Code Explorer and Code Inspections windows have spinning/busy icon with 0 open projects

[rubberduck-vba/Rubberduck] bclothier reopened issue #2780: Code Explorer and Code Inspections windows have spinning/busy icon with 0 open projects

Comintern

@Duga Too soon. That's in #4717.

this

yeah, need to go to bed. Just want to fix one more bug.

Duga

[rubberduck-vba/Rubberduck] comintern created comment on issue #4597: Windows "busy" spinner continues to spin

> #4717 addresses the worst of this, but there is a second issue (beside the one mentioned above) that will set the cursor to it's busy state. If the navigate command changes the code pane, the cursor will stick in the busy state until the mouse is moved. This promises to be incredibly difficult to fix in that we don't have direct access to the Cursor from inside the navigate command and it's asynchronous.

I'm inclined to ignore that last remaining bit in that the workaround is to simply mo

Comintern

Yep, I should probably get to bed too. The low hanging fruit is just so tantalizing though...

Duga

[rubberduck-vba/Rubberduck] build for commit 17f43348 on unknown branch: AppVeyor build failed

BUILD FAILURE!

Comintern

6:46 AM

Wut?

The sort broke 2 tests? Seriously?

this

@Comintern IKR?

Comintern

I suspect a sequence equity check that doesn't belong there...

A'yup.

    Assert.IsTrue(config.UserSettings.CodeInspectionSettings.CodeInspections.SequenceEqual(
            viewModel.InspectionSettings.SourceCollection.OfType<CodeInspectionSetting>()));

Duga

[rubberduck-vba/Rubberduck] comintern synchronized pull request #4717: Inspection Result window performance overhaul (Round 1)

[rubberduck-vba/Rubberduck] comintern commented on Rubberduck.Core/UI/Inspections/InspectionResultsViewModel.cs of pull request #4717: Inspection Result window performance overhaul (Round 1)

[rubberduck-vba/Rubberduck] rubberduck-vba-releasebot published prerelease Prerelease-v2.3.1.4445

[rubberduck-vba/Rubberduck] rubberduck-vba-releasebot created tag Prerelease-v2.3.1.4445

[rubberduck-vba/Rubberduck] build for commit 81a26265 on next: AppVeyor build succeeded

[rubberduck-vba/Rubberduck] bclothier created comment on issue #4625: Return value not used false positive given object variable

> The issue is with this part:

https://github.com/rubberduck-vba/Rubberduck/blob/b4f45bf0f076519375fc3a0467d1974c919477af/Rubberduck.CodeAnalysis/Inspections/Concrete/FunctionReturnValueNotUsedInspection.cs#L107-L120

The argument list comes up `null` which cause the procedure to return `true`, which I think is incorrect.

It's not exactly clear to me why we're doing this, though - it *looks* like this was meant to handle the case where the return value is used as an argument to some oth

Duga

7:21 AM

[rubberduck-vba/Rubberduck] build for commit a5d15bc1 on unknown branch: AppVeyor build succeeded

Duga

8:22 AM

[rubberduck-vba/Rubberduck] MDoerner created comment on issue #4625: Return value not used false positive given object variable

> The actual problem here is that this code makes the naive assumption that just because the reference is in a call statement the call is on the referenced function.

To fix this, the tree probably has to be traversed upwards stepwise

[rubberduck-vba/Rubberduck] MDoerner assigned MDoerner to issue #4719: Code Inspection: Incorrect recommendation to switch from ByRef to ByVal for array

M.Doerner

@this I saw that you have assigned yourself to the array ByRef issue. I have that inspection already completely rewritten in a branch.

Btw, I introduced the QualifiedModuleName member on the Declaration as a shortcut at some point because we need the module far more often than the qualified member name.

M.Doerner

8:41 AM

@Comintern Why would anything ever query a COM object for the component type when it already had the declaration? The component type is cached on the QMN.

Duga

11:40 AM

[rubberduck-vba/Rubberduck] Greedquest created comment on issue #4722: Crash upon VBE resize

> @bclothier Sure, where's it located?

this

1:50 PM

@M.Doerner oh ok - I've been trying to fix small bugs in my branch. It only is a unit test and minimal changes. If you intend to add your branch soonish, I can discard the changes before I open a PR on my bug fixes branch.

Duga

[rubberduck-vba/Rubberduck] bclothier created comment on issue #4722: Crash upon VBE resize

> It would be in this path:

`HKEY_CURRENT_USER\Software\Microsoft\VBA\X.0\Common` where `X` is either `6` or `7`, most likely `7` since you are running Excel 2016.

[rubberduck-vba/Rubberduck] bclothier edited comment on issue #4722: Crash upon VBE resize

> It would be in this path:

`HKEY_CURRENT_USER\Software\Microsoft\VBA\X.0\Common` where `X` is either `6` or `7`, most likely `7` since you are running Excel 2016. It's probably best to export the entire subkey just in case it's not just the `Dock` setting.

Simon Forsberg

@MathieuGuindon stars > open issues

Tweeting Duck

Unexpected but awesome interview from @SwiftOnSecurity, who, btw, is one of the best accounts around https://twitter.com/SwiftOnSecurity/status/1086446451323031552

Tweeted by pptsapper on January 19, 2019 at 4:03 AM

Mathieu Guindon

2:18 PM

@SimonForsberg haha good point!

Simon Forsberg

@TweetingDuck You lost your picture? If anyone finds a new picture, let me know and I'll fix it

Duga

[rubberduck-vba/Rubberduck] bclothier created comment on issue #4625: Return value not used false positive given object variable

> So for this snippet `GetIt(1).Select`, I have this parse tree:

`usage.Context` => `SimpleNameExprContext`, `33..33`
`.Parent` => `IndexExprContext`, `33..36`
`.Parent` => `MemberAccessExprContext`, `33..38`
`.Parent` => `CallStmtContext`, `33..38`

from there, the parent is the `MainBlockStmtContext`.

I'm not sure if I want to simply check if the call statement is its immediate parent or draw a list of acceptable contexts before going to the call statement.

We can see that:
``

Vogel612

@SimonForsberg found it :)

Simon Forsberg

@Vogel612 Thanks

Duga

[rubberduck-vba/Rubberduck] bclothier created comment on issue #4722: Crash upon VBE resize

> Possibly related: #4607

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

> @daFreeMan are you still experiencing this?

Duga

2:41 PM

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

> I am. I'll be sure to grab a couple of crash logs when I'm back in the office.
I'm getting crashes from Excel when I've got several workbooks open, not just on exit now. I'm not convinced it's an RD issue, but I feel that I'll get a better and more responsive investigation from you guys than I ever would from MS. ;)

[rubberduck-vba/Rubberduck] daFreeMan created comment on issue #4645: False positive 'Parameter can be passed by value'

> I'll have to run a quick test on the most recent build. I'll let you know what I find.

[rubberduck-vba/Rubberduck] daFreeMan created comment on issue #4597: Windows "busy" spinner continues to spin

> I'm inclined to agree with you. Once the user is aware of the situation, it's no big deal at all. However, for the uninitiated, it does appear that RD is "locked up" until you try to move the mouse. I'm impressed that you were able to find and fix some of it! Thanks.

[rubberduck-vba/Rubberduck] djrivers opened issue #4726: Folders multiply every time RD is refreshed

> Every time you try to refresh RD, folders multiply themselves producing empty folders.

![folder multiplication issue](https://user-images.githubusercontent.com/40927759/51428115-b6269580-1bc5-11e9-837f-36465dcd24d3.png)

Also, it won't allow you or won't see folder structure/annotation deeper than two levels.

![folder layers](https://user-images.githubusercontent.com/40927759/51428244-f1759400-1bc6-11e9-9eb6-48a1eade0dce.png)

Version 2.3.1.4414 is not showing any of these issues.

Mathieu Guindon

3:02 PM

@Duga yikes

this

I think @Comintern let in some tribbles on board.

Duga

[rubberduck-vba/Rubberduck] bclothier created comment on issue #4625: Return value not used false positive given object variable

> Actually, fortunately, we had 2 other unit tests that covers a related scenario which helped narrow down the correct changes:

`FunctionReturnValueNotUsed_ReturnsResult_InterfaceMember` and `GivenInterfaceImplementationMember_ReturnsNoResult`

Unfortunately, I'm not 100% happy with the changes which currently passes all the tests including the new unit test reproducing this issue. This is the change:

```
var indexExpr = usage.Context.GetAncestor<VBAParser.IndexExprContext>(

Duga

3:44 PM

[rubberduck-vba/Rubberduck] Vogel612 added label regression to issue #4726: Folders multiply every time RD is refreshed

[rubberduck-vba/Rubberduck] Vogel612 added label bug to issue #4726: Folders multiply every time RD is refreshed

[rubberduck-vba/Rubberduck] Vogel612 added label critical to issue #4726: Folders multiply every time RD is refreshed

[rubberduck-vba/Rubberduck] Vogel612 added label feature-code-explorer to issue #4726: Folders multiply every time RD is refreshed

[rubberduck-vba/Rubberduck] Vogel612 created project card

[rubberduck-vba/Rubberduck] Vogel612 milestoned issue #4726: Folders multiply every time RD is refreshed

IvenBach

in Coding Projects and Factorio Heaven, 1 hour ago, by Simon Forsberg

Oh geez... https://www.amsterdamduckstore.com/rubber-duck/trump-rubber-duck/

That’s some fowl and awful humor.

Duga

4:04 PM

[rubberduck-vba/Rubberduck] bclothier created comment on issue #4557: Square-bracket expressions are flagged as Global variables

> A band-aid would be to exclude the `BracketedExpression` declaration type from the inspection's query.

However, it's a bit deeper than that. That code wouldn't be legal in other VBA project for where there are no `appobject` that has an evaluate member on its interface. Furthermore, we probably don't want to exclude other uses of bracketed expressions.

```
Public Sub derp()
Dim foo As Range
foo = [A1] 'runtime expression
foo = Application.[A1] 'VBAProject.Mo

[rubberduck-vba/Rubberduck] DaveSemon created comment on issue #4711: RD Loads but does not function - Stays at PENDING

> Well, after days of trying various fixes, I have to give up. I cannot get RD to work. I removed all add-ins, reinstalled RD - still did not work. I uninstalled MS Access and all add-ins, then reloaded Access and RD. Same problem. And, I did this on 2 computers. So, thanks for all the help, but .... I'll try again in a couple of months.

Comintern

@M.Doerner That's a really good question. I'm going to have to dig up the place I found it.

Duga

[rubberduck-vba/Rubberduck] comintern assigned comintern to issue #4726: Folders multiply every time RD is refreshed

this

4:20 PM

@Duga bizarre.... @IvenBach you can install and run RD, right?

Comintern

@Duga I need to figure out HTH to unit test that...

Huh? Why would I be able to get 2 assert dialogs open at the same time originating from the UI thread?

Duga

[rubberduck-vba/Rubberduck] Greedquest created comment on issue #4722: Crash upon VBE resize

> That's the whole lot saved as a text file
[VBA entry.txt](https://github.com/rubberduck-vba/Rubberduck/files/2775734/VBA.entry.txt)

I believe the VBE got in that state by crashing in some other way (I would never make the window that small on purpose). I've been messing around with the rubberduck api recently if that has any bearing

[rubberduck-vba/Rubberduck] bclothier created comment on issue #4711: RD Loads but does not function - Stays at PENDING

> That's sad to hear and frustrating as we don't have a reproduction on our end. It may be related to the need for shimming (c.f. #4041 ) - we have observed somehow similar error with #3837 though the symptoms are quite different (this has to do with loading RD whereas you can load it just fine but then error on the casting the commandbars).

That said, because you are on an Insider build, it can't hurt to report this, especially if you have a 3rd computer that isn't on a insider build and ca

Comintern

4:40 PM

@Duga I did notice in the log file that the resize event is hitting the message pump before the window is subclassed. Wasn't Vogel's resize hack in the WndProc?

Duga

[rubberduck-vba/Rubberduck] bclothier created comment on issue #4711: RD Loads but does not function - Stays at PENDING

> One more question that may seem totally random -- do you have or have you ever installed Office version older than 2003? (2000, 97...)

this

@Comintern No, but might be indirect

github.com/rubberduck-vba/Rubberduck/pull/3866/files/…

Comintern

@this That gets invoked from SubClassProc. The WM_SIZE is hitting the pump before we're in the message queue.

Oh wait, maybe not - I don't think the path that takes is logging anymore.

this

4:59 PM

I thought it worked because the weird parameter had special meaning to VBE so it had to go through. It would be us who need to not tale the "special" message

Comintern

Yeah, it would be there to prevent passing a bad size down to the managed control.

Do our toolwindows load in a deterministic order? It's specifically the 4th one that's the problem.

this

Not sure. Might have to do with reading registry

Comintern

Yay, hexdump!

this

ikr?

Comintern

I'm tempted to drop the contents into my registry...

I wonder if I can just copy and paste that into HxD...

I seriously do not want to type that.

M.Doerner

5:14 PM

@Duga I start to get the impression that we really should not have merged the CE changes before the release.

Now, we are stuck with fixing all issues introduced before we can release.

this

well, the release was slanted for the CE improvements. Without that, I'm not sure we have things to justify a new release? If need be, we can hold back release a bit more until all CE issues are fixed.

Comintern

That's my fault for assuming this was covered in the existing tests.

M.Doerner

The reference explorer is not a justification for a release?

this

5:30 PM

Oh dear. I thought that was released already :(

Comintern

Well, if we want to release clean with the reference explorer, the CE merge should be clean enough to roll back.

M.Doerner

No, that was the main feature of the upcoming 2.4 release.

Neither has managing attributes via annotations been released yet.

this

Can we merge master to a specific commit in next?

M.Doerner

Not taking the latest next will be non-trivial.

We still need the translations.

Comintern

6:30 PM

@this Were you still looking for instances where SCWs are in use at teardown?

2019-01-19 12:21:34.6122;TRACE-2.3.1.21944;Rubberduck._Extension;Initiating App.Shutdown...;
2019-01-19 12:21:34.6267;TRACE-2.3.1.21944;Rubberduck._Extension;Disposing IoC container...;
2019-01-19 12:21:34.6522;TRACE-2.3.1.21944;Rubberduck.UI.DockableToolwindowPresenter;Disposing DockableWindowPresenter of type Rubberduck.UI.UnitTesting.TestExplorerDockablePresenter.;
2019-01-19 12:21:34.6580;TRACE-2.3.1.21944;Rubberduck.UI.DockableToolwindowPresenter;Disposing DockableWindowPresenter of type Rubberduck.UI.ToDoItems.ToDoExplorerDockablePresenter.;

You get that when you shut down while Inspector.FindIssuesAsync is returning results.

M.Doerner

7:33 PM

The ParameterCanBeByValInspection is missing soo many tests.

Comintern

Speaking of tests, do you see any issue with using the MockParser to in a static ctor to generate a base list of declarations?

M.Doerner

Sorry, I fail to parse that sentence.

Comintern

I'd like to get a static List<Declaration> to use on bunch of related tests. I was thinking of using the MockParser in a static constructor.

It seems silly to repeatedly parse the same test project if all I need is the declaration list.

M.Doerner

You could statically cache the declarations, if they are all you need.

Alternatively, you could generate the declarations by hand.

Anyway, gtg.

Comintern

OK, that's what I thought. I'm thinking that will significantly improve the test runtimes.

this

8:14 PM

@Comintern that and other subsystems need to listen for Terminate event to cancel their operations

Comintern

That's kind of what I was thinking.

On the upside, the trouble with tribbles is sorted out. Needs a crapload more UTs though.

Duga

11:24 PM

[rubberduck-vba/Rubberduck] MDoerner created comment on issue #4625: Return value not used false positive given object variable

> What do you do in the case there is no IndexExpr in the chain because the function does not have parameters?

M.Doerner

11:41 PM

@Duga Moreover, whenever you handle index expressions in a special way, you also have to do it for whitespace index expressions.

Transcript for