VBA Rubberducking

Comintern

00:00

@IvenBach I think mine was installed with the Windows SDK. If you have VS installed, I'd just search your C drive for it.

ThunderFrame

or install OleWoo

Duga

RELOAD!

ThunderFrame

OleView seems to choke on some libraries

Duga

[Hosch250/CheckersUI] 1 commit. 46 additions. 3 deletions.

Comintern

Or put a breakpoint in the ReferencedDeclarationsCollector. :-)

Duga

00:01

[retailcoder/Rubberduck] 25 commits. 241438 additions. 112609 deletions.

[rubberduck-vba/Rubberduck] 35 commits. 13 issue comments. 71826 additions. 36537 deletions.

ThunderFrame

or install the future Rubberduck 3

Duga

[VSDiagnostics/VSDiagnostics] 1 opened issue.

IvenBach

@Comintern I'll check for it at home. Always curious to learn new things.

ThunderFrame

Build a better Object Browser #1093

IvenBach

What exactly IS a left function? github.com/rubberduck-vba/Rubberduck/issues/…

Comintern

00:04

Left("Foobar", 3)

ThunderFrame

Left and Left$ don't really exist in the TLB

IvenBach

@ThunderFrame Then what exists? I use left and I know it works. This explanation is going to give me a headache I bet.

Comintern

_b_var_Left and _b_str_Left

IvenBach

:notamused:

Comintern

The first returns a Variant, the second returns a String.

IvenBach

00:08

Why are they named they way they are?

Comintern

The $ at the end is a type hint. VBA uses the type hint to decide which function to call internally.

Mat's Mug

And the b stands for "basic"

Comintern

Huh. TIL.

IvenBach

@Mat'sMug ok

Comintern

I'd figured it was something dumb like "be a Variant, Left".

ThunderFrame

00:10

they're creations of VBA to alias the real (but hidden) functions like _B_var_LeftB

Mat's Mug

lol

@Comintern has to do with how BASIC strings aren't null-terminated like C strings I think

Comintern

Ah BSTR v CSTR.

Mat's Mug

Yeah

ThunderFrame

anyone care to guess what the result of this is?

Mat's Mug

aka proper Hungarian notation ;-)

ThunderFrame

00:13

?[_B_var_LeftB]("ABCD",3)

Comintern

That's why Len(string) is so fast. With a null terminated string it would require a full traversal of the string.

@ThunderFrame Segfault.

ThunderFrame

@Comintern Seriously ?[_B_var_LeftB]("ABCD",3) Returns "A"

Comintern

^Force of habit.

ThunderFrame

WTF?

Comintern

I just jump to the conclusion at a good portion of my code or your code has a decent probability of a segfault. I know for my code it's a good first guess.

ThunderFrame

00:17

no, I mean the result of the function

Comintern

Unicode.

ThunderFrame

paste it into you immediate window - it's like _B_var_LeftB is ignoring the second argument

Comintern

/0A/0B/0C/0D

Mat's Mug

You asked for the leftmost 3 bytes...

Comintern

^

ThunderFrame

00:17

ah

Mat's Mug

Left != LeftB ;-)

ThunderFrame

I thought the alias was using a B on the end, rather than it being an alias for LeftB

Comintern

And MidB is a freak'n statement too...

Stupid keyboard hook.

WTH did MS have to stop calling CallNextHookEx in Office 2013?

IvenBach

@ThunderFrame From this point on I got lost... Could one of you give me an explanation/implication of what that means?

Mat's Mug

ThunderFrame confused left and leftb

ThunderFrame

00:24

> A BSTR (Basic string or binary string) is a string data type that is used by COM, Automation, and Interop functions.

IvenBach

Is it just that the LeftB is dealing with the Bytes and you need to know how the computer stores the information to truly appreciate what's going on?

Mat's Mug

LeftB, RightB, MidB, LenB all deal in bytes

?LenB("a")

Should print 2

?Len("a")

Should print 1

Useful when reading/writing from/to binary files

I suppose

Never needed to use them as far as I recall

Although, they probably perform better than their no-B brothers

Comintern

LenB is more useful for Types

It's also useful for sizing byte arrays to pass to API functions.

Type Foo
    Bar As Long
    Baz As Long
End Type

Public Sub Example()
    Dim x As Foo
    Debug.Print LenB(x)  '<- 8
End Sub

Mat's Mug

Ah, yes!

Comintern

Public Function Is64Bit() As Boolean
    Static test As LongPtr
    Is64Bit = LenB(test) = 8
End Function

Mat's Mug

00:39

#If Win64 Is64Bit = True #Else Is64Bit = False #EndIf

^ should compile in vba6 ;-)

Comintern

There was some reason I used that once, although the why completely escapes me now.

ThunderFrame

So it looks like I'll be presenting on Rubberduck at the Microsoft Sydney campus in July.

3

Comintern

:thumbsup:

IvenBach

@Mat'sMug Where can the character representations of the letters be found it?

Comintern

www.ascii-code.com?

IvenBach

00:44

Would this be similar to using a Number falling within [65-90,97-122] inside of =Char(Number) on the worksheet?

Comintern

Exactly the same.

IvenBach

The blind pig finds another truffle!

That makes sense why all the values with LenB() were returning 2 with [a-zA-Z] and LenB("AA") was 4

Mat's Mug

@ThunderFrame huh that's awesome!

ThunderFrame

@Mat'sMug Financial Modeling Conference - ModelOff's Global Training Camp

Mat's Mug

01:20

> Rubberduck.Setup.2.0.12.0.exe (5.86 MiB) - Downloaded 100 times.
Last updated on 2017-03-05

IvenBach

@ThunderFrame I hope everything goes well. You've put a lot of work into it so far.

Even after seeing the sausage being made I'm committed to eating more.

3

Mat's Mug

01:45

@IvenBach do you know about regular expressions?

I ask because I'd like to walk you through the heart of Rubberduck - the VBAParser.g4 grammar; it helps if you know how regex works ;-)

IvenBach

@Mat'sMug I've used them and feel I have a basic understanding of them.

Don't use them all too often. As such I'm not extremely comfortable with them, yet.

Mat's Mug

not a problem - it's not like Rubberduck's parser is all regex (turns out it was, before we discovered Antlr)

IvenBach

I'm very interested in learning. It'll have to be once I get home. An hour or so and I should be home.

Mat's Mug

awesome

IvenBach

I hope my class explanation wasn't too horrible for @puzzlepiece87 earlier.

Mat's Mug

02:01

I thought it was fine :)

Duga

[rubberduck-vba/Rubberduck] retailcoder merged pull request #2810: Tests for 2737 from MDoerner/Rubberduck/TestsFor2737 into rubberduck-vba/Rubberduck/next

[rubberduck-vba/Rubberduck] retailcoder closed issue #2737: The DeclarationFinder cannot cope with multiple classes implementing the same interface.

IvenBach

That makes 1 of us. I know what I want to say, it makes sense to me, but when I'm explaining it to others I get blank stares.

Duga

[rubberduck-vba/Rubberduck] MDoerner pushed commit 5e30a0c5 to next: Fixed stupid error in the RubberduckParserState

[rubberduck-vba/Rubberduck] MDoerner pushed commit 923d97de to next: Some cleanup in the DeclarationFinder

Mat's Mug

@IvenBach I totally understand... I'm hoping I can explain RD's parser :)

Duga

[rubberduck-vba/Rubberduck] build for commit 280f29f2 on next: AppVeyor build succeeded

IvenBach

02:05

I'm sure you'll be able to explain it fine. The real question is whether or not I'll be able to understand what you're explaining.

Mat's Mug

I'm hoping I understand what I'm explaining lol

the grammar should go smoothly though

100+ downloads already for v2.0.12 - and with that today, all versions combined, Rubberduck hit 10,000+ downloads! Thanks for your support!

Tweeted by rubberduckvba on March 8, 2017 at 2:06 AM

great, first like is a "buy followers & likes" spam account lol

crap, what was I working on

makes a PR just to see the diff

Duga

[rubberduck-vba/Rubberduck] retailcoder closed issue #2805: Contextual status label should tell us the scope we're in

[rubberduck-vba/Rubberduck] retailcoder pushed 12 commits to next

[retailcoder/Rubberduck] MDoerner pushed commit 5e30a0c5 to rd-next: Fixed stupid error in the RubberduckParserState

[retailcoder/Rubberduck] MDoerner pushed commit 923d97de to rd-next: Some cleanup in the DeclarationFinder

Mat's Mug

@Duga huh?

Duga

[rubberduck-vba/Rubberduck] build for commit 817509e9 on next: AppVeyor build succeeded

[rubberduck-vba/Rubberduck] retailcoder reopened issue #2805: Contextual status label should tell us the scope we're in

[rubberduck-vba/Rubberduck] retailcoder commented on issue #2805: Contextual status label should tell us the scope we're in

> Reopening. The referenced commit was sort-of-reverted, because it made context-sensitive selection description annoyingly flaky. Truth is, the FindSelectedDeclaration function needs a serious refactor.

[rubberduck-vba/Rubberduck] retailcoder added label status-declined to issue #2806: Inspection for Enum name with implied plural

[rubberduck-vba/Rubberduck] rubberduck203 commented on issue #2597: Concurrency issues randomly make tests fail

[rubberduck-vba/Rubberduck] retailcoder commented on issue #2806: Inspection for Enum name with implied plural

> Declining in its current form - also... I think we have something like 60 pending inspections requested/suggested but not implemented.

[rubberduck-vba/Rubberduck] retailcoder closed issue #2806: Inspection for Enum name with implied plural

FreeMan

02:25

@Mat'sMug You should schedule some "classes" for those of us who'd like to contribute but need some guidance.

Mat's Mug

@FreeMan that's a good idea. I'll save the conversation :)

chat.stackexchange.com/rooms/info/14929/…

Duga

[rubberduck-vba/Rubberduck] comintern commented on issue #2597: Concurrency issues randomly make tests fail

> @rubberduck203 It's 5 tests, all related to binding in the resolver. Usually about 2 or 3 will fail on my home machine and 1 or 2 on my work machine. It's never the same ones, and they always pass (at least for me) if they're run by themselves.

[rubberduck-vba/Rubberduck] retailcoder commented on issue #2597: Concurrency issues randomly make tests fail

> @comintern I've had an instance or two of one (or two) failing more than once. like, had to run it individually 3 times and it turned green. Slightly annoying issue!

[rubberduck-vba/Rubberduck] rubberduck203 commented on issue #2597: Concurrency issues randomly make tests fail

> That's actually very telling @comintern. Tests are cross talking. Multi-threading was mentioned. Is async-await involved here or something lower level?

Any chance you guys can link me to one of the tests?

FreeMan

@Mat'sMug umm... I don't see it. At least not yet.

Duga

02:41

[rubberduck-vba/Rubberduck] comintern commented on issue #2597: Concurrency issues randomly make tests fail

> @rubberduck203 EnclosingProjectComesBeforeOtherModuleInEnclosingProject is a good example. The multi-threading would be in the parser - IIR it's all just straight-up parallel threads. I don't recall any asych-await.

Hosch250

02:53

@Mat'sMug What would you think if you could write C# inline in WPF XAML docs?

Mat's Mug

yuck?

Hosch250

Kind of like you can do in a cshtml doc in ASP.NET.

So, you could inline converters where they make sense.

I don't mean to that extent, just a little bit.

Mat's Mug

you could always put them in the code-behind... which is a partial class

Hosch250

Oh, a sub-class?

I never thought of that.

Mat's Mug

I think if people can inline C# in XAML markup, it's going to get abused like there's no tomorrow

Hosch250

02:55

Except, I'm not sure if you could bind them properly that way.

Probably :/

I tried out the new C# features today, except for pattern matching.

You can't just use the tuples thing--you have to reference System.ValueTuple first.

Mat's Mug

makes sense

Hosch250

You can't just use the tuples thing--you have to reference System.ValueTuple first.

Mat's Mug

you just said that

so it's more of a framework thing than a language thing

Comintern

I'd much prefer the opposite direction - making it easier to emit xaml from the code behind.

Hosch250

I only typed it once.

Comintern

02:59

So would xaml inlining be like string interpolation?

Hosch250

Kind of.

Comintern

I avoid that like the plague. When you have functionality encoded into string literals it makes it a serious PITA to refactor.

Hosch250

Well, no, it would actually be compiled, like in the .cshtml files.

Comintern

var foo = "Foo's bar baz = {bar.Baz}"; doesn't play nice with thing like find all references.

It breaks at run-time, not compile time.

Hosch250

I don't get what in the world you are talking about.

I was talking about something like:

Mat's Mug

03:03

@Comintern R# pays no mind

Hosch250

<TextBox Text="{Binding Foo, Converter=@{C# code here}} />

Mat's Mug

VS either actually

Hosch250

Then, the tool that compiles the XAML into C# will just make that a converter automatically.

Mat's Mug

I don't think you need converters that often.. there's almost always a better solution that doesn't involve one

Hosch250

Like, introducing an intermediate variable in the VM?

Comintern

03:05

Try building the interpolated strings at runtime. That came up in a code review at work a couple months ago.

ThunderFrame

eek, inline C# in XAML sounds like ASP Classic

Hosch250

@Comintern I have no idea what you are talking about.

ThunderFrame

downloads FrontPage.NET

Comintern

lol

ThunderFrame

@Mat'sMug I take it there haven't been any requests for RD support in FrontPage VBE?

Comintern

03:07

ASP.RD?

Mat's Mug

FrontPage was a VBE host?

ThunderFrame

Which MS intern thought adding an IDE to the FrontPage IDE was a good idea

Duga

[rubberduck-vba/Rubberduck] rubberduck203 commented on issue #2597: Concurrency issues randomly make tests fail

ThunderFrame

@Mat'sMug yep, I think so

Duga

> Thanks. Man, I miss VS right about now. I don't see anything obviously off, but *anytime* I've ever run across an issue where a test ran in isolation, but not many at once, one of two things was going on.

1. The test was inadvertently keeping global state (this one isn't).
2. A test was running a "fire and forget" asynchronous method.

I'm theorizing that something is being spun off into the background, but the test doesn't know to wait for it. I don't see anything that would be doing

Mat's Mug

03:08

gosh... I suppose it "just works" then

ThunderFrame

msdn.microsoft.com/en-us/library/office/…

Duga

[rubberduck-vba/Rubberduck] rubberduck203 commented on issue #2597: Concurrency issues randomly make tests fail

> Thanks. Man, I miss VS right about now. I don't see anything obviously off, but *anytime* I've ever run across an issue where a test ran in isolation, but not many at once, one of two things was going on.

1. The test was inadvertently keeping global state (this one isn't).
2. A test was running a "fire and forget" asynchronous method.

I'm theorizing that something is being spun off into the background, but the test doesn't know to wait for it. I don't see anything that would be doing

Comintern

I'd check, but I usually go out of the way to exclude FrontPage from my Office installs.

ThunderFrame

I know it doesn't sound like something I'd do, but back in the day, I worked out some strange combination of line_continuations and square brackets that made FrontPage more usable.

3

Mat's Mug

lol

Comintern

03:12

I think I might be forced to deal with RegisterHotKey.

I thought I could bypass the LL keyboard hook bug in Office with a debug hook, but there really isn't a good way to deal with the fact that a keypress generates 11 callbacks...

...and 11 more when you release the key.

Hosch250

Oh gosh, @Mat'sMug, how did you guys pull this off? twitter.com/rubberduckvba/status/837120748829491200

Something in the resolver?

IvenBach

I has returned

Mat's Mug

woot!

Hosch250

The parsing certainly couldn't get much faster, and the declarations finder usually goes really fast too.

The references resolver is (was) the slowest stage when I left...

Comintern

@Hosch250 90% of that is the declaration finder.

Mat's Mug

03:17

the inspections are the slowest thing now

Hosch250

What did you do?

Mat's Mug

[magic.gif]

Comintern

We cached all of the common lookups.

IvenBach

@Mat'sMug do you have time to review the parser like you mentioned earlier?

Hosch250

@Mat'sMug Are they in the parse pass now?

Mat's Mug

03:18

@Hosch250 not yet

Hosch250

Ah, interesting.

@Mat'sMug That will slow it down.

Comintern

How so?

The further inspections get from Declaration lists, the better.

Mat's Mug

@Hosch250 not much - it's one function call that's given a target and decides whether to tack an inspection result onto it

Hosch250

Calculating the inspections is what takes a while.

Although technically, we can cache the ones for modules that weren't changed.

That might be something we could look into.

Mat's Mug

@IvenBach got VS/RD open?

@Hosch250 exactly

Hosch250

03:20

I'll leave you to your fun--dishes time.

Mat's Mug

later!

Comintern

@Hosch250 Only because they are trawling lists of 10s of thousands of Declarations. The more information that the parser contexts hold, the less we need to do that.

Hosch250

Oh, spring vacation is next week.

IvenBach

I have VS open but need to update RD. How again to I get my version updateto the latest release?

Hosch250

I'm going to be fiercely busy though, anyway, so RD might have to wait until summer.

Mat's Mug

03:23

@IvenBach I usually click the repository link at the top-right of this screen, hit "new pull request", and PR [next] into my fork

IvenBach

That's in VS or on github?

Mat's Mug

on github :)

just above the tag cloud

fun pushing-it rubberduck rubberduck-programming vb6 vba whyskey-tango-foxtrot

IvenBach

At the compare changes screen

which is base and which is next?

Mat's Mug

the dropdown on the left is the "target"

the dropdown on the right is where it comes from

so left would be your fork, right the main repository

room topic changed to VBA Rubberducking: Discussions about general VBA. Also the war room of the Rubberduck project (rubberduckvba.com | github.com/rubberduck-vba/Rubberduck). [antlr] [com-interop] [parsing] [rubberduck] [vb6] [vba] [whiskey-tango-foxtrot]

Duga

[rubberduck-vba/Rubberduck] IvenBach created pull request #2811: Merge pull request #2797 from rubberduck-vba/next to merge master into next

> Release 2.0.12

Mat's Mug

03:28

now they're actual tags that link to CR questions :) ..except for one though

IvenBach

Did I get that correct?

Mat's Mug

looks like you put the main repo as the recipient

IvenBach

oops let me switch that

Duga

[rubberduck-vba/Rubberduck] IvenBach rejected pull request #2811: Merge pull request #2797 from rubberduck-vba/next

[rubberduck-vba/Rubberduck] build for commit 166f2e3f on master: AppVeyor build cancelled

IvenBach

so Left is my repository and the right is the main?

@Mat'sMug ^

Mat's Mug

03:30

yeah. left is where the commits go, right is where they come from

Duga

[rubberduck-vba/Rubberduck] IvenBach created pull request #2812: Merge pull request #2797 from rubberduck-vba/next to merge master into next

> Release 2.0.12

IvenBach

continuous-integration/appveyor/pr — AppVeyor build cancelled

I'm getting an error

This branch is out-of-date with the base branch
Merge the latest changes from next into this branch.

Mat's Mug

same

huh

IvenBach

I don't think my initial errors were ever resolved

Mat's Mug

base (left) is your fork, head (right) is the rubberduck-vba/Rubberuck repo?

IvenBach

03:33

Yep, at least I do believe so.

Mat's Mug

Duga

[rubberduck-vba/Rubberduck] build for commit 166f2e3f on master: AppVeyor build succeeded

IvenBach

So it just succeeded?

Mat's Mug

@Duga is only reporting what's happening on the main repo

well, unless you setup the webhook on your fork too

everytime a PR is created, AppVeyor schedules a build for it

if the PR breaks tests, the build will fail

IvenBach

do I do the pull request from my page github.com/IvenBach/Rubberduck or from RD page github.com/rubberduck-vba/Rubberduck?

Mat's Mug

03:37

yours

github redirects to the fork you pick as a base

IvenBach

That's probably why then. Retrying from mine

Mat's Mug

feel free to close PR #2 then ;-)

Duga

[rubberduck-vba/Rubberduck] IvenBach rejected pull request #2812: Merge pull request #2797 from rubberduck-vba/next

IvenBach

@Mat'sMug Makes a lot more sense now.

Mat's Mug

hi @PeterMTaylor!

IvenBach

03:40

I set it up like the image but it becomes
There isn’t anything to compare.

Peter M Taylor

Ah I'm caught browsing! ;)

Howdy to all

IvenBach

Didn't seem to like me switching the requests around.

@PeterMTaylor I'm struggling to be taught by @Mat'sMug!

Mat's Mug

@IvenBach github.com/IvenBach/Rubberduck/pull/2

Peter M Taylor

I must say jolly good chap! You are learning well ;)

IvenBach

@Mat'sMug Close that one, correct?

or Merge pull request?

Mat's Mug

03:43

eh, merge it - you'll make plenty of PR's anyway

IvenBach

Alright. That seems to have succeeded.

Mat's Mug

woot!

ok, so.. I'm looking at VBAParser.g4 in Rubberduck.Parsing/Grammar

IvenBach

So I now download the zip from my repo and extract it to me local folder for VS?

Mat's Mug

clone it

you can do that from within VS I think

IvenBach

Team Explorer>{Sync|Fetch|Pull}?

Mat's Mug

03:47

yeah

IvenBach

Do i choose Sync or Fetch or Pull?

Mat's Mug

wait is the team explorer connected to your local already?

pull

IvenBach

From what I recall Team Explorer is connected to my local but that one hasn't been updated

Mat's Mug

pulling will do that

are you in the [next] branch?

IvenBach

Yes

Mat's Mug

03:49

did you pull?

IvenBach

Clicked it and 'Cannot pull because there are uncommitted changes. Commit or undo your changes before pulling again. See the Output window for details.' came up

Mat's Mug

ok so you have local changes that haven't been committed yet - you need to either undo or commit them before you can pull the remote

IvenBach

how do I go about undoing since they won't be what I want

Mat's Mug

in the changes tab, right-click each file.. IIRC you can right-click the namespace too

select "undo"

then pull the remote changes in

IvenBach

Repository updated to commit 5882d5e5.

Mat's Mug

03:56

:+1:

ok. so bring up VBALexer.g4

IvenBach

Just for my knowledge. If a change is made in the repo I'm trying to pull from, and I make a change to my own local files. I'm not able to successfully do a pull till that conflict is resolved?

Comintern

9 times out of 10 you won't see a conflict.

Mat's Mug

it's not necessarily a conflict - your local changes must at least be committed before you can merge the remote stuff in

IvenBach

And what does committing my changes entail?

All this is completely new to me so bear with me.

Mat's Mug

that's fine - it means you're committing these changes to your repository.. they're "officially in" if you will

any modification you make to the code can be undone like you just did in the team explorer

IvenBach

03:59

and committing isn't to my local but to my GitHub repo?

Mat's Mug

commits are local

Transcript for