VBA Rubberducking

Duga

12:00 AM

RELOAD!

[Minesweeper] New Users: 9, Games Played: 68, Bombs Used: 53, Moves Performed: 9246

RELOAD!

[rubberduck-vba/Rubberduck] 2 issues closed. 3 issues opened. 20 issue comments

Afonso Mira

12:20 AM

Helloi

how can i see the translation in previow

Afonso Mira

12:47 AM

@M.Doerner Can you help me?

Mathieu Guindon

2:08 AM

@AfonsoMira is PT listed in the settings dialog? A while ago we had to modify the code for a new language to show up there, but looking at that code it looks like no changes are needed anymore, we just detect what cultures are available.

Duga

4:09 AM

[rubberduck-vba/Rubberduck] bclothier created comment on issue #5351: Unit Tests does not create a test module

> From what I see in the EnsureProjectReferencesUnitTesting, it does not attempt to add by GUID (e.g. References.AddFromGuid). It only uses the file (References.AddFromFile). We alrea

IvenBach

7:11 AM

in Coding Projects and Twitter Heaven :), yesterday, by Marc-Andre

https://blog.aurynn.com/2015/12/16-contempt-culture

From Simon’s room but that is a good read.

Really illustrative about what RD is doing for those that don’t have another environment to develop in.

RD is leaving your code better than it found it.

Vogel612

7:47 AM

@MathieuGuindon we might need a display name for the language...

Which is ... a not-so-great ux now that I think of it

this

9:31 AM

@Vogel612 FWIW, the installer already asks to select a language. That choice ought to carry over into the installed version, so that it shouldn't be necessary to go there; it would be there for benefits of those who want to switch language at runtime.

@MathieuGuindon Found a blog article that also implies that it's a normal thing to do

Duga

10:00 AM

[rubberduck-vba/Rubberduck] MDoerner created comment on issue #5351: Unit Tests does not create a test module

> If the value from the registry is indeed the description, we can simply change to FindByDescription.

this

10:47 AM

@MathieuGuindon turns out that you can run VBA on Commodore!

Son Goku ssj4

3:02 PM

Hi folks, nice weekend :-)

I've noticed there is a new 'Function return value gets discarded' inspection.

Is this a right behaviour? Or false possitive?

I've got function: 'vbNewLines'

Public Function vbNewLines(Optional ByVal numOfLines As Long = 1) As String
    Dim i As Long
    For i = 0 To numOfLines - 1
        vbNewLines = vbNewLines & vbNewLine
    Next i
End Function

When used in context:

Debug.Print "First line" & vbNewLines(2) & "Third line"

This raises the inspection.

The function can't be a procedure, because it's returning string which I then use. So... ? :-)

Vogel612

yea, that shouldn't raise the inspection result, actually

Function Return Value gets discarded is "just" a rewording of "ReturnValueNotUsed" or whatever it was before with some bugfixes rolled into there

Mathieu Guindon

4:00 PM

That's debug.print being annoying as usual... we do handle referencing expressions in argument lists, but Debug.Print doesn't have an argument list (like all Print statements) and we always forget to special-case it

Carlos Gustavo Reetz

Hi folks. There's something I've been wanting to ask for a while related to VBA. It's not a pressing issue, just something I observed, was unable to find references online, and I'm quite curious about.

If one runs multiple instances of a VBA Application, each instance will be able to execute code independently from the other, as long as there are hardware resources.

But it both instances try, at the same time, to use CreateObject (or even New), the one who gets to be the first will block the other until object creation is finished. For "light" objects its barely noticeable. But if one's t…

Hosch250

One more Linux thing I find annoying is that they don't really have a "select word by double-click" feature, apparently. In Windows, double-clicking in the FF addressbar would select just the specific path section bounded by special characters. That doesn't happen in Linux.

Hosch250

4:20 PM

Also, it doesn't know how to play gifs :(

Mathieu Guindon

@CarlosGustavoReetz CreateObject hits the registry and uses Win32 to spawn the object... so that's one shared resource (registry); then the Application object being created will initialize, and that initialization very likely accesses shared resources as well... this isn't a VBA thing, you'd have the same constraints spawning Application instances with any language.

Carlos Gustavo Reetz

@MathieuGuindon, Thanks! That would be my next question (whether it was a OLE/COM thing, or would manifest in .NET as well)

this

5:12 PM

@MathieuGuindon To be pedantic, the object creation is actually a part of the COM specifications, and has less to do with Win32. Normally, I believe the CreateObject internally makes use of the IClassFactory and the related library functions described in the 2nd method.

Registry lookup to find the CLSID usually should be quick and thus insignificant. I'd look more at what the application.

Mathieu Guindon

Yeah

this

@CarlosGustavoReetz Because you didn't mention which, it bears reminding that not all applications will have same behavior WRT the creation. Outlook is a good example; you can call CreateObject all you want, even up to 1000 times if you wanted but it only have a single instance so it actually acts like GetObject for the next 999 times.

So Mat's right - this isn't something to do with languages but how the application you're creating is implemented. They may have decided that creation requires exclusive access which cause the serialized access.

Duga

[rubberduck-vba/Rubberduck] SmileyFtW created comment on issue #5351: Unit Tests does not create a test module

> Since y'all are working on deeper issues with this I will leave closing this up to y'all. I am up and functioning again, so that is good enough for me... Thanks!

IvenBach

@SonGokussj4 thinking in air code of you Dim an array and then Redim it with the number of elements you want you could use the Join function to concatenate it together. Trust avoids creating and dumping the string in the loop.

M.Doerner

@SonGokussj4 That bug is on me. However, please note that this is not about whether the function can be written as a procedure. That is FunctionReturnValueAlwaysDiscarded, which will not fire for recursive functions..

Son Goku ssj4

5:27 PM

@IvenBach Thanks for the tip. I'll add that as TODO. :-) I'm not using this function much so it has not any performance impact. But good to know though :-)

M.Doerner

@MathieuGuindon You are right. I forgot that we are not only in an argument if there is a argument list ancestor, but also if there is an output list ancestor.

this

@M.Doerner could we write an ancestor function that consolidate all 3 usages? Mat said we always forget to special case, so if that's the case, it shouldn't have to be?

M.Doerner

That would probably always be forgotten. It would not conform to the usual GetAncestor<TargetContectType>().

this

Yeah, i guess it'd probably be more of FindUsages than a subtype of GetAncestor

M.Doerner

The question really is not if it is a usage.

The question is whether a reference is a procedure call.

There is a specific method for this in the inspection; it just lacks the test for output lists.

this

5:41 PM

Ok. I took Mat's comment to imply that there were other inspections that needed similar information.

IvenBach

@this in the docs you linked to it states ‘... getting a pointer to an interface...’. That means if you have Foo:Bar, Bar:Duk, and Duk:Bazz there are 3 interfaces that can be chosen for a Foo instance. 2 interfaces for Bar. Only a single interface for Bazz. Is my understanding accurate?

M.Doerner

I think it was more meant in the way that we tend to forget output lists rather frequently, because they are not something one has to deal with explicitly very often.

this

@IvenBach if they are exposed to COM, yes. Note that a object can implement additional interfaces that are not COM-visible, and by COM rules, you can have only one line of inheritance anyway

This means that Foo, Bar, Duk and Bazz each must be defined as a COM interface, and inheriting from the previous interface in order for your statement to be true.

IvenBach

Visibility being designated by the ComVisible attribute.

this

That's a .NET thing.

In C++ you make it COM-visible by writing an IDL for it and exposing that via the type library.

IvenBach

5:55 PM

Not knowing C++ is limiting my ability to fully understand.

IDL is Intermediate Disassembly Language?

this

Do not forget that COM is meant to be language agnostic. The ComVisible is just a .NET's shorthand to do the equivalent for you without you having to roll up your sleeves and writing a IDL

No, interface definition language.

docs.microsoft.com/en-us/windows/win32/midl/midl-start-page

Carlos Gustavo Reetz

@this indeed, I was mostly referring to Excel, though I remember to have tested it with Word and PowerPoint as well (Didn't have Outlook nor Access at the time).

Duga

[rubberduck-vba/Rubberduck] BZngr created comment on issue #3371: Refactor|Rename doesn't do anything on name collision

> Perhaps something changed in the `RenameRefactoring` since this issue opened. In v2.5 I can `RenameRefactor` `pFoo` to `Foo` successfully starting from the midpoint step and get the expected result described above. But, I'm thinking that renaming `pFoo` to `Foo` should only rename the single, local, (undeclared) instance of `pFoo` (whichever was selected) and leave all others untouched.

So...Although this issue can be closed as not reproducable, I'm wondering if there is a different iss

this

Hmm. Creating a new Excel instance is possible; not sure why it'd be blocking.

Are you just creating instances, and not also opening a particular file?

Carlos Gustavo Reetz

I was trying to come up with a "general purpose multi-process implementation library" to be used in VBA... (mostly for the fun of it)

this

6:08 PM

Ah. Cool. VBScript would have been cheaper, I think.

Carlos Gustavo Reetz

So I would have a Mastar Application that would spawn Worker Applications... to make that asynchronous (non-blobking) the spawning was being done via VBScript

that was when I noticed I wasn't able to spawn multiple instances at the same time... even firing multiple scripts simultaneously

each script would report back to the Mastar Application via GetObject (it had Public Functions written at the ThisWorkbook module) and hand over the newly spawned instances, when those were ready.

Duga

[rubberduck-vba/Rubberduck] codecov[bot\] created comment on pull request #5362: Fix FunctionReturnValue(Always)DiscardedInspection

> # [Codecov](https://codecov.io/gh/rubberduck-vba/Rubberduck/pull/5362?src=pr&el=h1) Report
> :exclamation: No coverage uploaded for pull request head (`FixFunctionReturnValueDiscardedInspection@c6f25a4`). [Click here to learn what that means](https://docs.codecov.io/docs/error-reference#section-missing-head-commit).
> The diff coverage is `n/a`.

Carlos Gustavo Reetz

At first I considered just using the pure instances, without creating workbooks, since I just wanted separated VBA Runtimes to be available... I also remember investigating whether it there was a "lighter" VBA host among Excel, Word and PowerPoint. But in the end I stuck with Excel... and also created a workbook along each instance, since it enabled the use of certain other useful functions (evaluate, worksheetfunctions and the like)

Duga

[rubberduck-vba/Rubberduck] retailcoder created comment on issue #3371: Refactor|Rename doesn't do anything on name collision

> Yeah the name validation / clash-verification logic was revised at some point leading up to 2.5.0; this one is probably a dupe of an issue about the clash-verification logic that was closed with the merge of a PR, leaving this one open.

Not sure I'm following the local `pFoo` problem - assuming a parse happened, I'm not seeing how the resolver could possibly treat these local undeclared variables as anything other than undeclared locals (does the toolbar not say something to that effect? or

[rubberduck-vba/Rubberduck] MDoerner added label resolver to issue #3371: Refactor|Rename doesn't do anything on name collision

[rubberduck-vba/Rubberduck] MDoerner created comment on issue #3371: Refactor|Rename doesn't do anything on name collision

> The new problem is a bug in the resolver. When we resolve undeclared variables, we reuse already found instances of the variable. Unfortunately, the check is by the qualified name of the enclosing procedure, which is identical for Let, Set and Get. We will have to split that according to declaration type.

[rubberduck-vba/Rubberduck] build for commit c6f25a4e on unknown branch: AppVeyor build succeeded

[rubberduck-vba/Rubberduck] codecov[bot\] edited comment on pull request #5362: Fix FunctionReturnValue(Always)DiscardedInspection

> # [Codecov](https://codecov.io/gh/rubberduck-vba/Rubberduck/pull/5362?src=pr&el=h1) Report
> Merging [#5362](https://codecov.io/gh/rubberduck-vba/Rubberduck/pull/5362?src=pr&el=desc) into [next](https://codecov.io/gh/rubberduck-vba/Rubberduck/commit/5d7369a3c8ed8df1848bd3583c0fb6a0e97887b7?src=pr&el=desc) will **decrease** coverage by `<.01%`.
> The diff coverage is `100%`.

```diff
@@ Coverage Diff @@
## next #5362 +/- ##
=========================

[rubberduck-vba/Rubberduck] build for commit c6f25a4e on unknown branch: 61.69% (target 0%)

[rubberduck-vba/Rubberduck] build for commit c6f25a4e on unknown branch: 100% of diff hit (target 60%)

Carlos Gustavo Reetz

6:38 PM

@this Oh, now I remembered. Each new instance would open a read-only copy of the Master Workbook, so they'd have the same set of code available to execution. To send something to be executed by a Worker, the Master App would, first, send over the function name, parameters and an optional Callback Object, and then Call a method that relied on Application.OnTime to actually trigger the Worker's calculations.

It was somewhat overengineered, but quite fun to watch it working

Duga

7:17 PM

[rubberduck-vba/Rubberduck] codecov[bot\] created comment on pull request #5363: Generate distinct unresolved variable declarations in Get Set and Let

> # [Codecov](https://codecov.io/gh/rubberduck-vba/Rubberduck/pull/5363?src=pr&el=h1) Report
> :exclamation: No coverage uploaded for pull request head (`FixUndeclaredVariableScopingInResolver@58e8e15`). [Click here to learn what that means](https://docs.codecov.io/docs/error-reference#section-missing-head-commit).
> The diff coverage is `n/a`.

Duga

7:33 PM

[rubberduck-vba/Rubberduck] build for commit 58e8e158 on unknown branch: AppVeyor build succeeded

[rubberduck-vba/Rubberduck] codecov[bot\] edited comment on pull request #5363: Generate distinct unresolved variable declarations in Get Set and Let

> # [Codecov](https://codecov.io/gh/rubberduck-vba/Rubberduck/pull/5363?src=pr&el=h1) Report
> Merging [#5363](https://codecov.io/gh/rubberduck-vba/Rubberduck/pull/5363?src=pr&el=desc) into [next](https://codecov.io/gh/rubberduck-vba/Rubberduck/commit/5d7369a3c8ed8df1848bd3583c0fb6a0e97887b7?src=pr&el=desc) will **increase** coverage by `<.01%`.
> The diff coverage is `70%`.

```diff
@@ Coverage Diff @@
## next #5363 +/- ##
==========================

[rubberduck-vba/Rubberduck] build for commit 58e8e158 on unknown branch: 61.69% (target 0%)

[rubberduck-vba/Rubberduck] build for commit 58e8e158 on unknown branch: 70% of diff hit (target 60%)

Duga

7:48 PM

[rubberduck-vba/Rubberduck] retailcoder closed issue #3371: Refactor|Rename doesn't do anything on name collision

[rubberduck-vba/Rubberduck] MDoerner pushed commit 58e8e158 to next: Generate distinct unresolved variable declarations in Get Set and Let

[rubberduck-vba/Rubberduck] web-flow pushed commit f27427f1 to next

Merge pull request #5363 from MDoerner/FixUndeclaredVariableScopingInResolver

Generate distinct unresolved variable declarations in Get Set and Let

[rubberduck-vba/Rubberduck] MDoerner pushed commit c6f25a4e to next: Make FunctionReturnValueDiscarded inspections correctly deal with output lists

[rubberduck-vba/Rubberduck] web-flow pushed commit 3ef11e38 to next

Merge pull request #5362 from MDoerner/FixFunctionReturnValueDiscardedInspection

Handle output lists (Print statements), fixes FunctionReturnValue(Always)DiscardedInspection

[rubberduck-vba/Rubberduck] retailcoder added label technical-debt to pull request #5361: One Style to rule them all, One Style to find them, One Style to bring them all and in the darkness bind them

[rubberduck-vba/Rubberduck] rubberduck-vba-releasebot prereleased draft prerelease Prerelease-v2.5.0.5303

[rubberduck-vba/Rubberduck] rubberduck-vba-releasebot created draft prerelease Prerelease-v2.5.0.5303

[rubberduck-vba/Rubberduck] build for commit f27427f1 on next: AppVeyor build succeeded

[rubberduck-vba/Rubberduck] build for commit f27427f1 on next: 61.69% (target 0%)

[rubberduck-vba/Rubberduck] rubberduck-vba-releasebot published prerelease Prerelease-v2.5.0.5303

[rubberduck-vba/Rubberduck] rubberduck-vba-releasebot created tag Prerelease-v2.5.0.5303

[rubberduck-vba/Rubberduck] build for commit f27427f1 on next: 70% of diff hit (target 60%)

Duga

8:14 PM

[rubberduck-vba/Rubberduck] web-flow pushed commit 9536b3ce to next: Update README.md

[rubberduck-vba/Rubberduck] rubberduck-vba-releasebot created draft prerelease Prerelease-v2.5.0.5304

[rubberduck-vba/Rubberduck] rubberduck-vba-releasebot prereleased draft prerelease Prerelease-v2.5.0.5304

[rubberduck-vba/Rubberduck] rubberduck-vba-releasebot published prerelease Prerelease-v2.5.0.5304

[rubberduck-vba/Rubberduck] rubberduck-vba-releasebot created tag Prerelease-v2.5.0.5304

[rubberduck-vba/Rubberduck] build for commit 3ef11e38 on next: AppVeyor build succeeded

[rubberduck-vba/Rubberduck] build for commit 3ef11e38 on next: 61.7% (target 0%)

[rubberduck-vba/Rubberduck] build for commit 3ef11e38 on next: 100% of diff hit (target 60%)

M.Doerner

9:32 PM

@MathieuGuindon I would appreciate if you could also take a look at my short PR from yesterday, i.e. #5360.

Mathieu Guindon

@M.Doerner yep, was giving AppVeyor a little break =)

M.Doerner

I think that will be useful for PR #5357.

Thx

Mathieu Guindon

yeah

M.Doerner

I have the feeling that my other larger PR is conflicting again.

Duga

[rubberduck-vba/Rubberduck] MDoerner pushed commit caa16668 to next: Let AddComponentService optionally take a component name

[rubberduck-vba/Rubberduck] web-flow pushed commit 98115229 to next

Merge pull request #5360 from MDoerner/EnhanceAddComponentToProvideTheName

Let AddComponentService optionally take a component name

M.Doerner

9:37 PM

Hm, apparently not.

I'll run the tests locally anywway to be sure.

this

Regarding the sheet access using string inspection - wouldn't using ActiveWorkbook implicitly be also a code smell?

Duga

[rubberduck-vba/Rubberduck] rubberduck-vba-releasebot prereleased draft prerelease Prerelease-v2.5.0.5305

[rubberduck-vba/Rubberduck] rubberduck-vba-releasebot published prerelease Prerelease-v2.5.0.5305

[rubberduck-vba/Rubberduck] rubberduck-vba-releasebot created draft prerelease Prerelease-v2.5.0.5305

[rubberduck-vba/Rubberduck] rubberduck-vba-releasebot created tag Prerelease-v2.5.0.5305

[rubberduck-vba/Rubberduck] build for commit 98115229 on next: AppVeyor build succeeded

[rubberduck-vba/Rubberduck] build for commit 98115229 on next: 61.69% (target 0%)

[rubberduck-vba/Rubberduck] build for commit 98115229 on next: Coverage not affected when comparing caa1666...9811522

M.Doerner

10:28 PM

@this There is an inspection for that, AFAIK.

Duga

[rubberduck-vba/Rubberduck] MDoerner synchronized pull request #5347: Let reference inspections use base class

[rubberduck-vba/Rubberduck] codecov[bot\] edited comment on pull request #5347: Let reference inspections use base class

> # [Codecov](https://codecov.io/gh/rubberduck-vba/Rubberduck/pull/5347?src=pr&el=h1) Report
> Merging [#5347](https://codecov.io/gh/rubberduck-vba/Rubberduck/pull/5347?src=pr&el=desc) into [next](https://codecov.io/gh/rubberduck-vba/Rubberduck/commit/999a846fec9e1b61cafe9a95b3c074ca0fb66702?src=pr&el=desc) will **not change** coverage.
> The diff coverage is `n/a`.

```diff
@@ Coverage Diff @@
## next #5347 +/- ##
=======================================

[rubberduck-vba/Rubberduck] build for commit 999a846f on unknown branch: Coverage not affected when comparing 999a846...999a846

M.Doerner

Yep, ImplicitActiveWorkbookReferenceInspection.

Duga

10:45 PM

[rubberduck-vba/Rubberduck] build for commit 18774ce2 on unknown branch: AppVeyor build succeeded

[rubberduck-vba/Rubberduck] codecov[bot\] edited comment on pull request #5347: Let reference inspections use base class

> # [Codecov](https://codecov.io/gh/rubberduck-vba/Rubberduck/pull/5347?src=pr&el=h1) Report
> Merging [#5347](https://codecov.io/gh/rubberduck-vba/Rubberduck/pull/5347?src=pr&el=desc) into [next](https://codecov.io/gh/rubberduck-vba/Rubberduck/commit/999a846fec9e1b61cafe9a95b3c074ca0fb66702?src=pr&el=desc) will **increase** coverage by `0.01%`.
> The diff coverage is `81.25%`.

```diff
@@ Coverage Diff @@
## next #5347 +/- ##
=======================

[rubberduck-vba/Rubberduck] build for commit 18774ce2 on unknown branch: 81.25% of diff hit (target 60%)

[rubberduck-vba/Rubberduck] build for commit 18774ce2 on unknown branch: 61.67% (target 0%)

M.Doerner

11:44 PM

@IvenBach The AddComponentService should now allow you to to specify everything you need for the new component in your PR.

Transcript for