VBA Rubberducking

right.. yet we still re-localize the menu items when settings change. possible order changed and how the menus get created before the hotkeys are wired up

FreeMan

@MathieuGuindon musta been... I don't remember seeing them. I'd have probably complained if I noticed them disappear...

Mathieu Guindon

pretty sure you did ;-)

FreeMan

musta been a loooooooooooooooooooooooong time ago, then!

Comintern

I'm not finding the IHotkey interface being used anywhere UI related though.

this

7:26 PM

IIRC, Rkapka changed how it worked

it's all driven in the settings, I think?

mansellan

@MathieuGuindon fwiw if I ever get the toolbar setup thing finished, it should make it trivial to inject differing menu setups. At which point we could have a setting for context menus - using subfolder \ integrated.

Comintern

Yeah, but if we were showing them anywhere we'd need to be accessing the key information to display it.

> // need to hook hotkeys before we localize menus, to correctly display ShortcutTexts

Mathieu Guindon

yeah everything got flipped around, there's no notion of hotkeys in CommandMenuItem anymore

mansellan

one thing i'm not happy with is that our menus are not customisable with the built-in editor, as we create them new each time. iiuc that's because teardown issues?

Mathieu Guindon

@mansellan not sure, worth experimenting with

Hosch250

7:31 PM

I think it might be.

mansellan

@MathieuGuindon agreed

@Hosch250 not sure though if that was ever fully confirmed, or just (strongly?) suspected

Mathieu Guindon

seems the only place that knows what hotkey goes to what command, is the hotkey settings

so, menus need to know about hotkey settings - and the hotkey wiring order doesn't matter anymore

(as long as we still re-caption the menu items when (language || hotkey) settings change)

mansellan

@Comintern yeah saw that - not sure why though?

Comintern

The settings were\are horribly intertwined.

mansellan

ah ok

Mathieu Guindon

7:34 PM

@mansellan that comment dates back from when CommandMenuItem knew about hotkeys

mansellan

oh ok

Mathieu Guindon

@Comintern would be worth investigating an ambient context pattern - be it just for that

Comintern

Ambient context?

Mathieu Guindon

RDContext.Current.Settings.IndenterSettings -> gets you the current indenter settings

this

you know, when you enclose something in amber.

Comintern

7:36 PM

Like candle-light and stuff?

this

so someone can then make a park out of whatever duck you enclosed it in.

Mathieu Guindon

blogs.msdn.microsoft.com/ploeh/2007/07/23/ambient-context

Comintern

Oh wait that's "Ambiance context".

this

did they misspell "Ambulance" there?

Comintern

The main roadblock to RDContext.Current.Settings.IndenterSettings etc. is the settings window.

IIR, it was the stupid hotkeys that kept me from doing that years ago.

Mathieu Guindon

7:38 PM

RDContext.Current.ParserState could possibly help with the CE holding on to the parser state and leaking it though

Comintern

That could help with a lot of potential leaks.

this

hmm, wasn't it also held up as an anti-pattern?

Mathieu Guindon

not quite. depends

this

e.g. ASP.NET's Current caused a lot of grief with testability.

Comintern

By purists.

If you build the context container correctly, it doesn't have to impact testability.

Mathieu Guindon

7:40 PM

it's effectively global state, I get it. point is, we do have state that needs to be global

@Comintern that

this

Yeah, I agree. As long it's substiutable, no problem on my part.

mansellan

meh, just use a public static

;-)

this

something like public static IRubberduckContext Context, I guess.

mansellan

ttqw

Comintern

In fact it would probably make test setup way easier.

IMHO we completely underutilize IoC in the test suite.

this

7:43 PM

because we can now mock VBE's insanity?

which reminds me. need to get back to 4072.... Get it going

Mathieu Guindon

@Comintern not sure. tests run sequentially, right?

Comintern

They can.

Mathieu Guindon

the thread-safety in there needs to be A-1

Comintern

I shudder to think how long that would take though.

Mathieu Guindon

how do you not mangle the parser state when you have 200 concurrent tests constantly tweaking the RPS?

this

7:50 PM

aren't they mainly managed with locks?

Mathieu Guindon

it's one thing to have 1 IDE running concurrent parses against 1 context

it's another thing to have 200 (mock) IDEs doing the same against ....1 context

Comintern

@MathieuGuindon In the tests, it can return the parser state in transient context.

Mathieu Guindon

I guess that's what .Current tracks

Comintern

That's part of what I mean by "underutilize IoC in the test suite".

Mathieu Guindon

we do

FreeMan

8:09 PM

On parse, I think it would be a good idea to keep the CE ducks spinning during Resolving references. There's nothing RD related that can be done while that's happening, so they may as well.

Mathieu Guindon

the idea was to have the CE available as soon as it's usable

Resolving references means CE nodes can be navigated

Hosch250

You can navigate from it at that point, IIRC...

Mathieu Guindon

most commands are disabled until parser state is ready though

FreeMan

hmmm... I guess I can dbl-click a method to navigate, but usually I'm renaming 1/2 a dozen things in a row, so I'm finding it a bit frustrating.

Maybe it's just my use case at the moment...

I see the ducks stop spinning and hit ctrl-shift-r and... nothing happens. then D'oh!

Mathieu Guindon

watch the parser state label instead of the spinning ducks?

FreeMan

8:15 PM

Yeah, but I often pop in here while it's parsing. My browser is only about 1/2 screen, so I can see the ducks next to it. They stop, I alt-tab, ctrl-shift-r... then watch the label for bit.

I guess it's just me.

it often is...

this

hmm. if it was made a bit less opaque to indicate background processing, would that help?

FreeMan

made what less opaque?

this

right now it's blue, right?

reach the Resolved references stage, then make it almost transparent blue

then at Ready, spinning ducks genocide.

But TBH, I'm a bit ambivalent on that.

FreeMan

don't kill the duckies, just... shoo them away

Mathieu Guindon

@this lol

IvenBach

8:17 PM

@FreeMan Won't someone please think of the ducklings?!

Mathieu Guindon

easter egg: correctly click all duckies in sequence, and they fall to the bottom of the toolwindow

this

since a background processing indicator usually has to be out of the way so you can interact with the CE.

FreeMan

It's ok, it's prolly just me.

Mathieu Guindon

@this needs another IsBusy state for that

Duga

[rubberduck-vba/Rubberduck] MDoerner created comment on issue #4629: SCP inserting extra "(" in method signatures

> This has been fixed before. This is probably a regression introduces by the fix for Call statements.

this

8:18 PM

yeah. the label itself is kind already doing that job.

FreeMan

suppose it could be like Excel 2016 when code is running:

this

?

Mathieu Guindon

lol

this

surely you're not holding this up as a paragon of UX....

Mathieu Guindon

@Duga indeed. and worrying that no test failed for it

FreeMan

8:20 PM

@this oh, absolutely! It's the way MS does it, it must be right! Right?

this

More whacking for you, Mat.

Mathieu Guindon

starting to wonder if any of the SCP tests are worth anything

@FreeMan riiiiight

this

@FreeMan I guess so. they're the bright sparks that gave us ribbons.

Mathieu Guindon

I've come to love the ribbons

this

@MathieuGuindon FWIW I'm sure you wouldn't get as far had you kept with the original untestable implementation.

Duga

8:21 PM

[rubberduck-vba/Rubberduck] mansellan edited wiki page Installing

this

Meh. Ribbon is just a lateral move. Not exactly better, not exactly worse. Just different.

FreeMan

Sadly, it used to work great in 2010. Window stayed properly lit, Application.StatusBar got refreshed as expected (in colors or colours) that could be read. It was pretty great. then... "upgrade"

mansellan

@Duga Added high-DPI instructions

FreeMan

@MathieuGuindon still ugh

Mathieu Guindon

@this yeah... except I would expect a red-then-green test to turn back red when something breaks... would be nice.

this

8:23 PM

yeah agreed. I guess you need more tests because IINM, the behavior could be different given a different sequence of inputs.

FreeMan

At least there's the option to unpin the ribbon and get my screen real estate back. Then, it's pretty much like a "menu" gasp!. Sadly 99% of the people I work with use laptops with low resolution screens and refuse to hide the ribbon, even after I show them how much screen they get back, so they end up seeing like 4 rows of their spreadsheet...

@MathieuGuindon is that how tests are supposed to work???? Who knew!

:)

M.Doerner

Regarding the SCP tests, it is simply that the assumptions about what the VBE does are wrong.

Whether it swallows pairs of parentheses at the end of a line depends on the context.

Mathieu Guindon

some tests are taking shortcuts, too

yeah

M.Doerner

In a call statement it does, in a method declarations, it doesn't.

Mathieu Guindon

every test that isn't passing an original, prettified, and reprettified code strings, is cutting corners

M.Doerner

8:33 PM

Actually, it even adds them in a method declaration.

So, in a method declaration, you get the SCP simply by reprettifying and swallowing the keypress.

Mathieu Guindon

I kinda hate that tests need to specify how the VBE will prettify and re-prettify the original code, ...otoh I can't quite think of a way around it.

Duga

[rubberduck-vba/Rubberduck] MDoerner synchronized pull request #4535: Cache user COM projects - WIP

[rubberduck-vba/Rubberduck] build for commit e0d7df7b on unknown branch: AppVeyor build failed

BUILD FAILURE!

Mathieu Guindon

"prettify" -> the original code, prior to adding/processing the autocode
"reprettify" -> the prettified code, after adding/processing the autocode

M.Doerner

chocolatey is getting on my nerves.

Mathieu Guindon

rebuilding

Duga

8:41 PM

[rubberduck-vba/Rubberduck] weslleywang starred us

3

Mathieu Guindon

woot! 70 more and we hit the jackpot!

this

hmm. it might be possible to get rid of the poo-lately entirely.

Duga

[rubberduck-vba/Rubberduck] build for commit e0d7df7b on unknown branch: AppVeyor build failed

BUILD FAILURE!

Mathieu Guindon

ffs

FreeMan

They need to rename NuGet to PeanutButter, then Chocolatey and PeanutButter need to merge...

:)

this

8:43 PM

it looks like there are nuget packages already

OpenCover, CodeCov, and InnoSetup

we might be able to bypass the poo-lately entirely that way.

The only uncertainty is the OpenCover; we're asking for OpenCover.Portable but I don't see a exact corresponding nuget package.

Duga

[rubberduck-vba/Rubberduck] build for commit e0d7df7b on unknown branch: AppVeyor build failed

BUILD FAILURE!

[rubberduck-vba/Rubberduck] build for commit e0d7df7b on unknown branch: AppVeyor build failed

Comintern

Does this look like a wrapper leak to anyone else?

Duga

BUILD FAILURE!

Comintern

    return
        _state.AllUserDeclarations.SingleOrDefault(d =>
                d.DeclarationType == DeclarationType.Project && d.IdentifierName == Vbe.ActiveVBProject.Name);

this

i think so yes. Need to double check inside the ActiveVBProject definition

Mathieu Guindon

8:49 PM

Is it just me or @chocolateynuget isn't working on @appveyor anymore? It says it can't find the @InnoSetup package, getting 403/forbidden. Any ideas? Rebuilding PR for the 3rd time...

Tweeted by rubberduckvba on December 10, 2018 at 8:47 PM

this

but I'm pretty sure that's going to return a new SCW for VBProject in which case will leak.

Comintern

Yup, it's leaking.

M.Doerner

Just get the name prior to the LINQ statement.

this

stick your finger in there and hang on tight.

Comintern

FML, leaking all over:

private Declaration GetTarget()
{
    if (Vbe.SelectedVBComponent == null)
    {
        return
            _state.AllUserDeclarations.SingleOrDefault(d =>
                    d.DeclarationType == DeclarationType.Project && d.IdentifierName == Vbe.ActiveVBProject.Name);
    }

    return _state.AllUserDeclarations.SingleOrDefault(
            t => t.IdentifierName == Vbe.SelectedVBComponent.Name &&
                    t.ProjectId == Vbe.ActiveVBProject.ProjectId &&
                    new[]
                        {

M.Doerner

8:51 PM

Anyway, that must be really old; it is using `AllUserDeclarations in a stupidly inefficient way.

Comintern

Does the analyzer not find chained access in anon functions?

M.Doerner

Where is this?

FreeMan

@Comintern sieve of eratosthenes?

this

those aren't technically chained members

no, wait, it should be catching that.

Comintern

ProjectExplorerRefactorRenameCommand

Hosch250

8:52 PM

@FreeMan Sieve of Mr. MacGregor.

this

Is Vbe a property on the class?

Comintern

Sieve of Mr. Magoo.

@this It's a property of the base class.

this

hmm. that might be causing the analyzer to miss that

FreeMan

♫ Rabbit stew, rabbit stew... ♫

M.Doerner

Those types, aren't that all types with the Module flag?

Comintern

8:56 PM

If it isn't it probably should be.

FreeMan

Why is it so difficult for people to look at the calendar of all the invitees when scheduling a meeting in Outlook?

Mathieu Guindon

does anyone look at Outlook calendars?

or maintains them?

Hosch250

Me.

Mathieu Guindon

O__O

Hosch250

We use them at work.

@FreeMan Use the Scheduling Assistant.

It shows everybody's calendars right there with busy parts blocked off.

FreeMan

8:58 PM

Other Person: Hey can we meet on Friday?
Me: Sure, my Outlook calendar is up to date
OP: schedules a meeting on top of two other overlapping meetings
Me: _declines_ I've got a meeting until 2pm, we'll have to meet after that
OP: OK
Me: _looks for new meeting invite._

Mathieu Guindon

@Duga re-rebuilding

M.Doerner

I will have to look at them starting in March.

Duga

[rubberduck-vba/Rubberduck] build for commit e0d7df7b on unknown branch: AppVeyor build failed

BUILD FAILURE!

Mathieu Guindon

well that was fast

FreeMan

@Hosch250 yeah, I know how to do that, it doesn't seem that anyone else has a clue...

Comintern

8:59 PM

#FailFast

FreeMan

fail fast

this

@MathieuGuindon FWIW, it's at poo-lately server's side.

so kicking the build won't do anything

Mathieu Guindon

ok but AV is claiming all services green

FreeMan

@MathieuGuindon blame caching...

this

hmm given that it's giving 403, try and login to the poo-lately using the same credentials that AV'd be using

Mathieu Guindon

9:02 PM

like I know what AV is using.. some generated passphrase, encrypted once and never looked at ever again...

this

ah

M.Doerner

Thet method from above should probably look something like the following:

private Declaration GetTarget()
{
    string selectedComponentName;
    using (var selectedComponent = Vbe.SelectedVBComponent)
    {
        selectedComponentName = selectedComponent?.Name;
    }

    string activeProjectId;
    using (var activeProject = Vbe.ActiveVBProject)
    {
        activeProjectId = activeProject?.ProjectId;
    }

    if (activeProjectId == null)
    {
        return null;
    }

    if (selectedComponentName == null)
    {
        return _state.DeclarationFinder.UserDeclarations(DeclarationType.Project)

Comintern

@M.Doerner I'll go ahead and fix it.

Hosch250

Is the DaaS message important, or should I unpin it?

Mathieu Guindon

@rubberduckvba @chocolateynuget @InnoSetup This was side effect of datacenter move we started last week. We synchronized new IPs witch @chocolateynuget people and things should be good now. Sorry for the trouble.

Tweeted by appveyor on December 10, 2018 at 9:10 PM in reply to rubberduckvba

Hosch250

9:11 PM

> Sorry, that page doesn’t exist!

FreeMan

I'm slow... was just gonna say that

Mathieu Guindon

They deleted the tweet to fix a typo lol

this

it's still there?

or was there another typo besides witch?

Mathieu Guindon

the new tweet yes

LOL

yeah

this

i'm sure they did great in the speeling bee contests.

FreeMan

9:14 PM

^that

Hosch250

Might be voice-to-tweet.

@this I'd have done "spilling bee"

Oh. My. Gosh.

11

Q: Why is my D string showing a C♯ on my tuner

I was playing for a bit and didn't tune any string, but when I wanted to tune it in at the end of my session my tuner kept saying C♯ instead of D. Can anyone please tell me how to get my D string back to its original state?

LOL.

Mathieu Guindon

@Hosch250 BECAUSE YOU'RE NOT IN TUNE

saw that one yesterday, thought "play kazoo, it's simpler"

</evil>

Hosch250

LOL.

I really need to get one of those plastic trumpet things and prove they have more than one note.

Tweeting Duck

Is it just me or @chocolateynuget isn't working on @appveyor anymore? It says it can't find the @InnoSetup package, getting 403/forbidden. Any ideas? Rebuilding PR for the 3rd time...

Tweeted by rubberduckvba on December 10, 2018 at 8:47 PM

@appveyor 3rd build failed as well, status page all green, nothing changed since the last successful build.. Why the 403? https://ci.appveyor.com/project/rubberduck-vba/rubberduck/builds/20909440

Tweeted by rubberduckvba on December 10, 2018 at 9:07 PM in reply to rubberduckvba

this

9:29 PM

@MathieuGuindon consider this, though - if he managed to sing out Beethoven's 13th symphony on a kazoo, that's an incredible feat than if he did it on a guitar. ;-)

Hosch250

Beethoven's 13th symphony?

Mathieu Guindon

To be fair I tuned my guitars by ear for as long as I remember. Got a tuner, only used it to get the A tone (5th string), tuning would be too sharp if I did all strings as per the tuner. Might have something to do with the $15 price tag though

this

Whatever it was that he's famous for.

Hosch250

I used to be able to tune my siblings violins for them by ear. I played the clarinet--minimal tuning required.

this

I probably conflated that for 5th symphony with Mozart's, IDK.

Hosch250

9:33 PM

@this You mean the symphony with the chorus? That's the 9th.

this

whatever, that.

Hosch250

And the 5th is the da-da-da-DA one.

this

The point being - if he can play any symphony with only a kazoo, that's impressive, no?

Hosch250

Yes.

this

and thus making Mat look a chump in the process. ;-)

Hosch250

9:34 PM

Unless it's one of those post-modern ones.

this

Post modren anything => worthless

Mathieu Guindon

@this I'll be impressed with 3rd movement of Moonlight Sonata on kazoo

this

I don't care what they do, throwing a bucket on a canvas and calling it an art is not going to cut it for me. Likewise for banging on grate and calling it high form of music.

Mathieu Guindon

Ludwig van Beethoven - Moonlight Sonata - 3rd Movement

►

^ that's some serious mastery right there

Hosch250

I'm listening to the 5th on kazoo right now.

Duga

9:38 PM

[rubberduck-vba/Rubberduck] comintern synchronized pull request #4601: All your refs are belong to Duck

[rubberduck-vba/Rubberduck] comintern edited pull request #4601: All your refs are belong to Duck

Mathieu Guindon

I'm watching that vid on mute, and I can feel the song just by looking at the fingers dance

Comintern

I think that's everything but the unit tests.

FreeMan

^ ^^ those are 2 funny messages back-to-back

Mathieu Guindon

Hosch listening to 5th symphony on kazoo, Mat listening to a metal rendition of Moonlight Sonata?

Hosch250

No.

Mathieu Guindon

9:42 PM

ah, I'm missing the joke then :)

Hosch250

The video has all the features but the unit tests.

Mathieu Guindon

ah

Comintern

Would that be an integration test?

Hosch250

I'm not sure whether Beethoven would've smashed that guy's guitar over his head or loved it.

Comintern

He wouldn't have heard a thing.

Duga

9:52 PM

[rubberduck-vba/Rubberduck] MDoerner assigned MDoerner to issue #4626: @Description annotation

Mathieu Guindon

Mozart would probably have loved it, then wondered how that big loud box could be hooked up to his harpsichord

When this PR is merged, you'll have the list of referenced libraries in the Rubberduck Code Explorer, with an awesome new dialog to add/remove references. But yeah until then, it's Tools > References, then ...pain. — Mathieu Guindon 7 secs ago

this

10:09 PM

hmm. why did he decide to import it into vb6 today, though....

feels like a step backward, esp if he's going to have 64-bit office.

Duga

[rubberduck-vba/Rubberduck] build for commit 43d7092b on unknown branch: AppVeyor build succeeded

this

10:28 PM

@Comintern wondering - did you see / play with this before? easyhook.github.io/api/html/T_EasyHook_COMClassInfo.htm

Comintern

As a matter of fact, I did. I'd considered using it for RD Mocks.

this

Cool. Have you done any experimenting?

Comintern

Not yet. It's kind of been lurking on the radar though.

this

OK

Mathieu Guindon

10:42 PM

@Comintern RD ~~Rocks~~ ~~Flocks~~ ~~Blocks~~ Mocks

Comintern

Just one more way to mock PowerPoint.

Mathieu Guindon

LOL

Duga

[rubberduck-vba/Rubberduck] MDoerner opened issue #4630: MissingAttributeInspection needs a quickfix

> Currently, the `MissingAttributesInpsection` finds all instances where an annotation linked to an attribute is present in the code, but the corresponding attribute is missing. Unfortunately, there is no quickfix. So fixing the issue entails manually exporting the module, adding the attirbute, and then reimporting.

There should be a quickfix that adds the missing attribute.

[rubberduck-vba/Rubberduck] MDoerner added label feature-inspection-quickfixes to issue #4630: MissingAttributeInspection needs a quickfix

[rubberduck-vba/Rubberduck] MDoerner assigned MDoerner to issue #4630: MissingAttributeInspection needs a quickfix

[rubberduck-vba/Rubberduck] MDoerner added label enhancement to issue #4630: MissingAttributeInspection needs a quickfix

[rubberduck-vba/Rubberduck] retailcoder created comment on issue #4630: MissingAttributeInspection needs a quickfix

> Is the inspection only firing for module-level annotations? Because we can't have RD tweak member attributes, we lose them whenever we rewrite the module. Good idea for module-level ones though.

[rubberduck-vba/Rubberduck] retailcoder added label module-and-procedure-attributes to issue #4630: MissingAttributeInspection needs a quickfix

[rubberduck-vba/Rubberduck] MDoerner created comment on issue #4630: MissingAttributeInspection needs a quickfix

> I intend to add missing member attributes as well.
Fixing the removal of member attributes on rewrite will be in another PR.

M.Doerner

@Duga We have nearly everything in place to readd member attributes after a rewrite.

this

would that require multiple parses?

M.Doerner

10:49 PM

Yes, two.

Hosch250

But the one could happen entirely in the background?

Without affecting anything of the visible parse>

M.Doerner

We cannot readd them in the original rewrite.

Hosch250

Oh.

M.Doerner

The original rewrite that erases them is for the code pane code.

The attributes can only be added via the exported code.

Mathieu Guindon

@M.Doerner that would be awesome

M.Doerner

10:54 PM

Enabling this is one reason I introduced the IRewriteSession and IRewritingManager.

this

will the 2nd rewriting be done automatically?

FreeMan

Anybody noticing excessively sloooow page loads using Firefox? I'm on 63.0.3 64-bit and the last several days, it's been exceedingly slow to load stuff.

Even the management pages for my local server are painfully slow, so I know it's not my network connection

this

is anyone else getting an security alert dialog (like ones from IE6 days) about invalid SSL certificate when loading symbols in Visual Studio debuggers?

M.Doerner

Having the second parse start automatically was the plan.

Implementing the quickfix for the MissingAttributeInspection will bring me half the way to readding the attributes.

this

11:11 PM

Ok, that does concerns me because I know that I can cause parser errors if I try to work with code while parse is running in the background. Since parsing must run and we can't assume there are no user inputs, how do you ensure the state is not disordered again?

M.Doerner

Basically, I could check whether the code pane code is dirty, but then you will not get your attributes back, if you change the module during the first parse.

this

That makes sense. I suppose one could have a additional status indication on the status bar to warn that attributes aren't synchronized.

M.Doerner

Anyway, with the current implementation of the AttributesRewriteSession, the code pane for the module will close.

this

But at the point it's more important that the parser doesn't end up in a parser error state because I did not want to wait for it to finish.

M.Doerner

And it will not reopen, because that is not implemented yet.

this

11:18 PM

Good. closing & reopening code pane also makes it less likely that that one will get edited in midparse.

M.Doerner

The problem is that that only applies to the second parse, not the first.

What would be a good place in Rubberduck.Parsing for something adding attributes to members?

Unfortunately, this cannot go into Rubberduck.Refactorings because I need it in the RewritingManager to readd the attributes.

Peter M Taylor

11:38 PM

My wife passed me a joke...the oldest computer can be traced back to Adam and Eve...it’s was an a apple with extremely limited memory. Just one bite, everything crashed.

FreeMan

@PeterMTaylor LOL

Seems like it may have been an issue with memory. Resource Monitor was showing Memory Compression running at 25+% CPU, even though there was no indication that I was really out of memory. A reboot and MC is at 0%.

wonders if just restarting FF on occasion will "fix" it

Transcript for