VBA Rubberducking

FreeMan

18:01

Here are the complete properties from the working Connection Manager

and the broken ones

IvenBach

@23fc9a62-56de-47fb-97b4-737890 Figured it out myself. Thanks for helping me rubberduck through it.

FreeMan

The working DFC shows:

And the broken one shows this (can't get the hover over to work when VPN/RDC, but the error message is the A valid file name must be selected message I showed above)

M.Doerner

In case I have to change an inspection result text because the inspection now works differently, what should I do with the translations for languages I do not speak? Delete them?

FreeMan

nope, spewing my guts here didn't set off the Ah Ha!!! bell of finally seeing the forest, either. :(

23fc9a62-56de-47fb-97b4-737890

@M.Doerner Yes.

Unless the process changed in the last year or two.

FreeMan

18:13

@M.Doerner @Vogel612 since you're both here and Max just brought it up - your English is good enough that I often forget it's not your first language.

Vogel612

@M.Doerner alternatively ask the translators for the respective languages to provide an updated text

that'd be Mat for French and SonGokuSSJ4 for Czech

this

@puzzlepiece87 watch out for leaks, though.

@FreeMan sorry, got pulled away. will look when I can

FreeMan

what? you have actual work to do? I thought you were RD's private SQL Server tech support!

no worries, I'll be here beating my head on my desk. if you see randogoaigpd8fha;jkn f; that means I missed the desk and hit my keyboard instead.

this

yeah, tire fire here. :\

FreeMan

ah. They'll burn for a while no matter what you do. Might as well take on a couple of pleasurable tasks for now. ;)

this

18:26

lol

KySoto

18:45

my sad powershell face....

0

Q: I am trying to update the data source of an oledb data adapter and nothing happens

I am attempting to update data in an oledb datasource. I pull data from a table that who's schema has been written out as an access create table script (i could probably find a better way to get that schema information) and stored in the AccessCreateScript field of the lookup table. I am able g...

this

@FreeMan just checking - did you disable the validation check?

that might be the difference between the two

FreeMan

Sounds very familiar, can't see where to do it in the properties of which item, though. that's why I've been through them 10 bajillion times

this

two places- on the project itself and on the package

won't find them in the indidivual tasks

FreeMan

takes a deep breath. Dives in

Mathieu Guindon

@Vogel612 we have Spanish too now, no?

Vogel612

18:48

ohhh right

this

email alias for them?

just a thought

Mathieu Guindon

deleting the translation is fine IMO

it'll stick out like a sore thumb next round of translations

FreeMan

@this Ah! Package|Execution|Delay Validation one is True one is False. True works...

this

yeah you need true because you cna't validate the expressions

SSIS IS FUN, KIDS!!!!

FreeMan

ba-da-bing! That was it!

Mathieu Guindon

18:50

tbh delayvalidation set to false is a royal pita when you have hundreds of connections

FreeMan

My keyboard, desk and forehead all thank you!

this

^^^^^^^^^^^^^^^^ so that.

FreeMan

@MathieuGuindon will keep that in mind if/when I ever get beyond the 5 I've got now...

this

and frankly it's kind of stupid.

Mathieu Guindon

just opening the damn package takes forever

this

18:52

can you imagine the ducky going "a second, lemme parse this bad boy you got here"?

FreeMan

^takes zero imagination whatsoever... :/

Considers turning off inspection auto-run

this

In my case I do turn off inspections. :\

Duga

[rubberduck-vba/Rubberduck] IvenBach opened issue #5097: OLE dialog on running unit tests

> **Rubberduck version information**
Version 2.4.1.21157
OS: Microsoft Windows NT 6.1.7601 Service Pack 1, x64
Host Product: Microsoft Office 2013 x86
Host Version: 15.0.5023.1000
Host Executable: EXCEL.EXE

**Description**
A semi-frequent occurrence where a dialog, shown below, is presented before unit tests are run. This is not reproduce-able every time. ~5-10% of the time when running tests the dialog comes up. It locks up Excel for ~3-5 minutes before dialog is displayed. Once OK is

[rubberduck-vba/Rubberduck] IvenBach added label bug to issue #5097: OLE dialog on running unit tests

this

@IvenBach can you repro on any files?

I'm concerned that this might be due to a certain unit test in a certain file, hence my question.

M.Doerner

For the ImplicitDefaultMemberAssignmentInspection, should the result contain the expression the assignment goes to?

IvenBach

18:57

The only thing that's consistent in my PMW. It's not just on a specific file.

KySoto

man i hope that between reddit and SO i get some halp with my script

M.Doerner

Like Assignment to '{0}' implicitly assigns to the default member '{1}' of class '{2}'`

Or should we avoid having the expression in the result?

this

I like the idea of having expression - remember that inspections is also a teaching tool

Mathieu Guindon

^ I like it too

this

and for those unfamiliar with the problems, they might not realize it is a problem and why.

~~I wonder if we need the complete expression, too....~~ nah. they can click on it to go the line

Mathieu Guindon

19:00

how about "implicitly assigns to the default member '{1}.{2}'" where 1: the class and 2: the default member?

@this that's the job of the detailed description and xml-docs :)

this

Yes but you need a good hook

IvenBach

Can anyone else repro that TestExplorer CopyResultsCommand isn't working?

Mathieu Guindon

this

Good form, Smee! That's a good form!

M.Doerner

The nice thing about putting the expression there is that the text is close enough to the original that I can fix all but the Czech result texts.

Mathieu Guindon

19:05

@IvenBach works here (2.4.1.4848)

IvenBach

Hm...

Mathieu Guindon

output's StartTime and EndTime don't feel right, but it's pasting a table in Excel

IvenBach

CodeInspections works, just not TestExplorer

Mathieu Guindon

weird

this

Time for another coffee accident on your computer, methinks.

IvenBach

19:08

Well it'll paste into excel but not N++.

this

hmm are we generating plain text?

Mathieu Guindon

I don't think we are

this

then that'd be why

seems weird, though.

Mathieu Guindon

we do HTML though

ThunderFrame didn't feel like lining up ASCII-art borders, I guess :)

this

hmm. I think to work, we must also write to the text format, I think.

LOL

there is a nuget package for that!

Mathieu Guindon

19:10

lol, of course there is

this

but seriously, CSV format to the plain format of the clipboard should work.

IvenBach

Both paste to excel without issue. CodInpsections can and does paste to N++ while TestExplorer doesn't. AFAICT the code's the same.

Mathieu Guindon

hm

IvenBach

_clipboard.AppendString(DataFormats.UnicodeText, textResults); is commented out. Methinks I spotted the issue.

Mathieu Guindon

the CE doesn't export plain text either AFAIKT

this

19:11

Well, we don't do cod inspections, so you should contact the authors who made this.

Mathieu Guindon

cod imp sections

@IvenBach probably due to a bug though

but yeah CSV makes more sense than ASCII-art

this

@MathieuGuindon in that case, git blame's yer friend

IvenBach

At this point I'm poking it to see what happens.

this

yeah I wouldn't want ascii boxes. It's OK for a command line tool that shall remain nameless but not so much for sharing.

IvenBach

Got it to paste just fine.

Mathieu Guindon

19:15

@IvenBach can you confirm it also makes the CE export to text?

IvenBach

CE currently doesn't paste to N++. Want me to edit the code and see if it can?

I get the feeling this functionality should be consolidated to a single locale.

Mathieu Guindon

sure

YES

feel free to refactor as needed :)

IvenBach

Hrm... Seems that command isn't even set up in CE.

CodeExplorer, TestExplorer, CodeInspections any anywhere else this is used?

Mathieu Guindon

todos

(IIRC)

@IvenBach huh, it does export to Excel/html

mind you, CE has a few commands, it's probably just buried in the bunch

this

19:42

@MathieuGuindon talking of which --- I need to remind there's a PR about commands that needs reviewing....

may need @M.Doerner to review it as well since he has requested few changes and I want to be sure all requests has been addressed.

M.Doerner

I'll have a look once I have finished my current PR.

this

:+1:

puzzlepiece87

@this Uh oh. How should I watch out for leaks?

KySoto

gents... extremetech.com/extreme/…

Mathieu Guindon

got a phone call from the optometrist, I'm getting my new glasses after work, and contact lenses that are actually at my prescription (using approximate trial lenses rn, and crazy-glued my broken old glasses but.. they're really beyond done) - I'll see sharp again tonight!

M.Doerner

19:55

Question on Let coersion default member resolution: if the reolution fails, should I simply return the underlying expression?

this

to whom?

if to users, I think that's the best thing we can do

Mathieu Guindon

in what context would resolution fail?

M.Doerner

I currently return a FeiledReolutionExpression, but that breaks the ObjectAssignmentRequiresSetInspection.

If a default member with a value type is requires, but there is none.

Basically, when the inspection should create a result.

With the failed resolution, there is no assignment reference.

Mathieu Guindon

could we have an IllegalLetCoercionExpression?

(and would that make the inspection easier to reliably implement?)

M.Doerner

What would that buy over just returning the underlying expression?

The expressions are gone after the resolver did its job.

Mathieu Guindon

19:59

ah, ok

so.. if we return the underlying expression, we get the assignment IdentifierReference against the "wrong" declaration?

M.Doerner

yes

Mathieu Guindon

that sounds like a good idea

I think

M.Doerner

It will just be as if we did not reolve the default member call due to Let coercion.

this

I wonder if that needs to be sometihng else

e.g. a collection of FailReference as opposed to IdentifierReference

mainly to avoid mixing bad and good apples in one basket

Mathieu Guindon

or IllegalIdentifierReference : IdentifierReference?

23fc9a62-56de-47fb-97b4-737890

20:02

> Ahh.. this thing would be good for cooking my steak. 20x20 cm is big enough and you could adjust the temp by downclocking or overclocking. XD

LOL.

this

Yes that could work.

Mathieu Guindon

hm, I don't remember if it's a struct or a class

@M.Doerner basically the problem with returning the underlying expression is that we basically "lose" the knowledge of something being off with that expression, right?

M.Doerner

I think we are good with just using the original one. You will only get let assignments to object type variables, if they are illegal.

All legal ones go to the default member.

Mathieu Guindon

seems inspectable enough

FreeMan

@MathieuGuindon Smee, I've had an apostrophe!

this

20:05

i worry that might cause some tests to fail as that might change the assumptions about what identifier references we can expect to find for something.

Mathieu Guindon

@FreeMan LOL

FreeMan

@IvenBach I'm a Cod fish!

M.Doerner

I do not see how that can make tests fail. Using the original expression is what we currently always do.

Mathieu Guindon

@this IDK, there are definitely a number of places where we assume default member calls aren't resolving correctly - IMO tests now failing because of that need to be fixed, not pampered

M.Doerner

Returning the FailedResolutionExpression is actually what makes quite a few tests fail rightt now.

BTW, there are only two places with tests affected.

I already fixed the ImplicitDefaultMemberAssignmentInspection, which obviously could no longer work the old way.

The other place is the ObjectVariableNotSetInspection.

Mathieu Guindon

20:09

nice

this

:+1:

FreeMan

@MathieuGuindon #LateToTheParty as usual, but I do what i can

this

@FreeMan but if you're chopped up in pieces, does that make you a codspiece?

FreeMan

ouch

this

>:D Leave it to English to butcher the language up!

FreeMan

20:16

Potentially dumb question: Access form with 2 radio buttons. For button A, I've defined the Option Value to be 1, for button B, I've defined it as 2. When I reference Form_myForm.ButtonGroup.Value I get values of 0 and 1. Shouldn't they be 1 and 2?

Yes, yes it should, but you actually have to assign the value from the form to the class that you're looking at later, otherwise the property remains at the nifty default value of 0 that VBA assigns it.

:/

this

instead of using constants, use the option values.

e.g. myForm.ButtonGroup.Value = myForm.OptionOne.OptionValue

FreeMan

> I've defined the Option Value to be 1, for button B, I've defined it as 2

Oh, gotcha. I created an enum for them to refer to them later.

this

yeah, no need for an enum. Just use the OptionValue property

then you can use any old values you like and code still works™

FreeMan

hrm...

but.. later, when I need to reference it from a a class that's not the form, I need to know which option I'm working with. Don't I need some way to know which one's which?

this

I see

FreeMan

20:21

I mean, I guess I could do If Implementation_ButtonOption = myForm.ButtonOptionA.OptionValue Then, but that feels really... clunky...

instead I've got Implementation_ButtonOption = optionA (which, with the capital at the end looks like it's spelled backwards, but isn't), and that seems to read much better

this

TBH I don't use them that frequent, and certainly not outside a form. I can't remember if you can write to OptionValue but if you could, you could assign the Enum members to the option controls instead.

that way, the enum is now The One True Source Of Truth™.

FreeMan

looks into it

this

the example implies yes, it can be written at runtime

FreeMan

but, the enum is in the Interface, so I'd have to create an instance of the class at form load time just to read the enum to populate the numbers, and that seems a bit awkward, too...

then I'd have to throw away that instance because I might not even need it later.

Makes A Lot Of Sense™, but feels really awkward in this instance.

since we're ™ random phrases...

this

wait

why does enum have to be on the class?

Mathieu Guindon

20:28

@FreeMan huh??

FreeMan

it's what you guys helped me with yesterday.

Mathieu Guindon

a public enum's location makes no difference whatsoever IME.

FreeMan

I guess it doesn't matter where it's located...

this

^^ & ^

Mathieu Guindon

oh, "read" as in "read with one's actual eyes"

#fail

this

20:29

#nowConfusedAgain

M.Doerner

I think the following should really not return an inspection result.

Private Sub DoSomething()
    Dim target As Object
    target = CreateObject(""Scripting.Dictionary"")
    target(""foo"") = 42
End Sub

Mathieu Guindon

@FreeMan you don't need to open up the interface to lookup the enum members... you can simply drill down and expand the CE node for it

this

I thought we were talking about reading enums into option values.

M.Doerner

The tests asserts that i does, though.

this

is that for the implicit assignment?

FreeMan

20:30

@MathieuGuindon nope. read = reference.

Mathieu Guindon

@FreeMan ignore me then :)

FreeMan

@this that

M.Doerner

Yes

FreeMan

poor choice of words.

this

I thought it should be target(""foo"").Value = 42?

M.Doerner

20:30

The assignment is to an unbound member call.

FreeMan

Use on RHS to assign to another variable somewhere

Mathieu Guindon

but target = CreateObject(""Scripting.Dictionary"") is missing a Set keyword

this

@MathieuGuindon facepalm leave it to me to look at small details and miss the big picture

M.Doerner

At compile time target(""foo"") has type Variant and is unbound.

@MathieuGuindon Not really, that is also an assignment to an unbound member from an unbound member.

Mathieu Guindon

huh, what inspection is that test for?

M.Doerner

20:33

Whenever Object is in play, you have no idea whether there is a default member.

this

@M.Doerner I presume that code actually won't run.

M.Doerner

ObjectVariableRequiresSetInspection.

I think it would fail because there is no key "foo".

Mathieu Guindon

target = CreateObject("whatever") absolutely should spawn a result

this

^

Mathieu Guindon

I think the assignment under it is just a red herring, should be removed

M.Doerner

20:34

Why?

Mathieu Guindon

because target is declared As Object

M.Doerner

The name of the test is ObjectVariableNotSet_GivenIndexerObjectAccess_ReturnsResult.

But exactly because it is declared as Object, the assignmant might be legal.

this

because of default member, yes. I realized that too late.

Mathieu Guindon

you're talking about target(""foo"") = 42, I'm stuck on the missing Set keyword in the instruction just above that

M.Doerner

It could resolve to a default member call as runtime.

this

20:36

it's the fact that that we are creating an object that trips us up

M.Doerner

It will spawn a result for ImplicitDefaultMemberAssignment, though.

this

@MathieuGuindon imagine it as target = CreateObject("Excel.Application") -- instead of getting the object Excel.Application, you get a string Application

since it resolves to Excel.Application.Name (the default property)

M.Doerner

And once it exists also for ImplicitDefaultMemberAccess.

this

(and we basically threw the object instance away)

Mathieu Guindon

I see that, and because target is declared As Object, the let-coercion is definitely illegal

in any case, that test code is confusing

M.Doerner

20:39

No, it is only illegal, if the object at runtime does not have a parameterless Property Let default member.

Mathieu Guindon

it's illegal because it throws error 91 at run time, ....exactly what the inspection intends to flag

this

@MathieuGuindon I have to agree, though. I assume it was set up like that to test for an edge case against a false positive. Needs a comment to explain that yes it really means to do this crapatastic code.

M.Doerner

At compile time, you get a resolution to an unbound member of type Variant.

Mathieu Guindon

Sub test()
    Dim target As Object
    target = Application 'error 91
End Sub

Duga

[rubberduck-vba/Rubberduck] daFreeMan opened issue #5098: Public enum value is not updated in CE after a parse

> **Rubberduck version information**
The info below can be copy-paste-completed from the first lines of Rubberduck's log or the About box:

Rubberduck version [...]
Operating System: [...]
Host Product: [...]
Host Version: [...]
Host Executable: [...]

**Description**
I changed the value of the first item of a `Public Enum` and parsed. The _Code Explorer_ continued to show the original value

**To Reproduce**
Steps to reproduce the behavior:
1. Create a `Publ

[rubberduck-vba/Rubberduck] daFreeMan added label bug to issue #5098: Public enum value is not updated in CE after a parse

M.Doerner

20:42

You are assigning to Nothing there.

Mathieu Guindon

I'm not

I'm assigning to _Global.Application

and Set is required

M.Doerner

No, you assign Application to a potential default member of Nothing.

Duga

[rubberduck-vba/Rubberduck] daFreeMan edited issue #5098: Public enum value is not updated in CE after a parse

M.Doerner

Public Sub TestIt()
    Dim target As Object
    Set target = Sheet1.Range("A1")
    target = Application
End Sub

Works flawlessly

Mathieu Guindon

holy shit

M.Doerner

20:44

With Object, everything happens at runtime.

Duga

[rubberduck-vba/Rubberduck] daFreeMan edited issue #5098: Public enum value is not updated in CE after a parse

M.Doerner

That is why all the default member crazyness works for late-bound code.

FreeMan

@Duga Submitted on home machine (because that's what I'm getting the screen shots from). Edited from the Work computer to paste in the RD version info & add the Log. But no, I couldn't add the log from there, so I had to mail it to myself and edit from the Home computer to paste the log in.

Thanks funky work network restrictions.

this

@MathieuGuindon remember, that's the same guy who bought us Wasabi.

Mathieu Guindon

yeah, no, something's off

target runtime type is Object/Range at target = Application, so there's an implicit default member call on both RHS and LHS

M.Doerner

20:48

Exactly

Mathieu Guindon

are we saying in target = Application, that the error 91 is coming from LHS attempting to invoke a default member?

(which completely makes sense, in retrospect)

M.Doerner

Yes

Mathieu Guindon

damn

M.Doerner

That is why you get that stupid misleading error message.

Mathieu Guindon

how can we make the (an) inspection say "you want a Set keyword there" anyway?

this

20:50

FWIW, I think that not having the Set for any CreateObject assignment is a huge smell in itself, esp. when using implicit default member access.

M.Doerner

As I said, we currently get a result from ImplicitDefaultMemberAssigmnetInspection.

Mathieu Guindon

that makes sense, but we'll lose our noobs here

M.Doerner

This now also covers unbound default member assignments.

Mathieu Guindon

I was hoping there'd be a way to flag it and quickfix it with "Add 'Set' keyword"

M.Doerner

We could add a SuspiciousLetAssignmentInspection.

Mathieu Guindon

20:53

which would flag LetStmtContext where LHS involves an implicit default member call? I can buy that.

M.Doerner

Even if it is legal, this kind of assignment will probably not be intentional.

Mathieu Guindon

yeah

this

^ we really want to bring that to the foremost

KySoto

*sigh

Mathieu Guindon

and fixed by making the LHS default member call explicit

KySoto

20:54

its been 2 hours, narry a comment for clarification

Mathieu Guindon

(or adding Set)

KySoto

reposted on spiceworks (it site) too

Mathieu Guindon

I'm really liking this

IvenBach

Regarding the use of ClipboardWriter to _clipboard.AppendFoo(bar, duk); since that's repeated does it make sense to add an AppendInfo(foo, bar, duck, ...) to that interface?

M.Doerner

We actually already have the one for the default member assignment.

I was rather thinking about ones where Object is involved or where both sides have a default member access.

Mathieu Guindon

20:57

picks up Range("A1") = 42

yeah, if RHS isn't an object then SuspiciousLetAssignment can't fire

but ImplicitDefaultMemberAssignment is up

M.Doerner

I always forget, is that strongly typed to Range?

Mathieu Guindon

yup

M.Doerner

Then we get that from the ImplicitDefaultMemberAssignment already.

Mathieu Guindon

yeah

58 secs ago, by Mathieu Guindon

but ImplicitDefaultMemberAssignment is up

and we are actually able to evaluate that RHS is an object type at this point?

M.Doerner

We would actually get both for the suspicious ones.

23fc9a62-56de-47fb-97b4-737890

20:59

@IvenBach You mean a params value?

Transcript for